From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 03:24:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 03:24:07 -0500
Subject: [LLVMbugs] [Bug 4479] register scavenger assertion failure
In-Reply-To: <bug-4479-206@http.llvm.org/bugs/>
Message-ID: <200907010824.n618O7UG028897@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4479


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Evan Cheng <evan.cheng at apple.com>  2009-07-01 03:24:03 ---
This has been fixed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 11:05:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 11:05:53 -0500
Subject: [LLVMbugs] [Bug 4487] New: instcombine fails
Message-ID: <bug-4487-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4487

           Summary: instcombine fails
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: opt
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rafael.espindola at gmail.com
                CC: llvmbugs at cs.uiuc.edu


"opt -instcombine" fails on the following test:

--------------------------------------
define i32 @test(i64 %a) {
bb20:
        %t1 = bitcast i64 %a to <2 x i32>
        %t2 = bitcast i64 %a to <2 x i32>
        %t3 = xor <2 x i32> %t1, %t2
        %t4 = extractelement <2 x i32> %t3, i32 0
        ret i32 %t4
}
--------------------------------------


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 11:54:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 11:54:49 -0500
Subject: [LLVMbugs] [Bug 4482] Wrong codegen for calls on darwin ppc32
In-Reply-To: <bug-4482-206@http.llvm.org/bugs/>
Message-ID: <200907011654.n61GsnOr014986@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4482


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-07-01 11:54:45 ---
fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090629/079912.html

This bug was exposed when llvm-gcc started using available_externally for
inline functions.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 12:23:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 12:23:34 -0500
Subject: [LLVMbugs] [Bug 4486] llvm-ld: Assertion `writers == 0 && " Writer
 lock already acquired!"'
In-Reply-To: <bug-4486-206@http.llvm.org/bugs/>
Message-ID: <200907011723.n61HNYEx016079@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4486


Owen Anderson <resistor at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Owen Anderson <resistor at mac.com>  2009-07-01 12:23:30 ---
Fixed in r74618.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 12:58:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 12:58:40 -0500
Subject: [LLVMbugs] [Bug 3494] assertion thrown in codegen of debug info
	"Multiple main"
In-Reply-To: <bug-3494-206@http.llvm.org/bugs/>
Message-ID: <200907011758.n61Hwekx017742@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3494


devang.patel <devang.patel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|dpatel at apple.com            |
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #15 from devang.patel <devang.patel at gmail.com>  2009-07-01 12:58:36 ---
Thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 16:02:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 16:02:05 -0500
Subject: [LLVMbugs] [Bug 4482] Wrong codegen for calls on darwin ppc32
In-Reply-To: <bug-4482-206@http.llvm.org/bugs/>
Message-ID: <200907012102.n61L25bc024351@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4482


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #3 from Dale Johannesen <dalej at apple.com>  2009-07-01 16:02:02 ---
We are now generating $stub instead of $non_lazy_ptr, so the correctness
problem is fixed.  However, there is usually no need to be generating stubs. 
It appears that we always are regardless of any flags, whereas x86-32 gets it
right.  


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 16:42:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 16:42:10 -0500
Subject: [LLVMbugs] [Bug 4487] instcombine fails
In-Reply-To: <bug-4487-206@http.llvm.org/bugs/>
Message-ID: <200907012142.n61LgAAq025828@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4487


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Dan Gohman <gohman at apple.com>  2009-07-01 16:42:09 ---
This is fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090629/079950.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  1 18:21:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 1 Jul 2009 18:21:48 -0500
Subject: [LLVMbugs] [Bug 4482] Wrong codegen for calls on darwin ppc32
In-Reply-To: <bug-4482-206@http.llvm.org/bugs/>
Message-ID: <200907012321.n61NLmAu028978@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4482


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Chris Lattner <clattner at apple.com>  2009-07-01 18:21:46 ---
"Wrong codegen for calls on darwin ppc32" is fixed.  If you want an enhancement
request, please add it to PowerPC/README.txt.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 05:59:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 05:59:46 -0500
Subject: [LLVMbugs] [Bug 4493] New: Add command-line option to specify
 signedness of 'int' bit-fields.
Message-ID: <bug-4493-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4493

           Summary: Add command-line option to specify signedness of 'int'
                    bit-fields.
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Basic
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: zaffanella at cs.unipr.it
                CC: llvmbugs at cs.uiuc.edu


According to the C99 standard (6.7.2p5):
"[...] it is implementation-defined whether the specifier 'int' designates the
same type as 'signed int' or the same type as 'unsigned int'."

Similarly, in the C89 standard it is said:
"Each of the above comma-separated lists designates the same type,
except that for bit-field declarations, signed int (or signed ) may
differ from int (or no type specifiers)."

clang seems to be missing a language/target option to specify this
implementation-defined behavior.

This should be added and exposed by supporting the GNU -fsignedbitfields
-funsigned-bitfields command-line options.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 06:16:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 06:16:29 -0500
Subject: [LLVMbugs] [Bug 4494] New: gold: should have a flag to dump the
	module
Message-ID: <bug-4494-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4494

           Summary: gold: should have a flag to dump the module
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


When bugs are encountered during LTO, it is currently hard to debug, because
llvm-ld doesn't always reproduce the bug.

It would be useful to be able to dump the module after linking, or just before
linking.

For example this assertion doesn't reproduce with llvm-ld
 Assertion failed: (ATI != AbstractTypeMap.end() && "Abstract type not in
AbstractTypeMap?"), function MoveConstant.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 06:25:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 06:25:18 -0500
Subject: [LLVMbugs] [Bug 4495] New: instcombine fails
Message-ID: <bug-4495-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4495

           Summary: instcombine fails
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: opt
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rafael.espindola at gmail.com
                CC: llvmbugs at cs.uiuc.edu


opt -instcombine fails on the input

---------------------------------------------
define i32 @test(i64 %test) {
entry:
        %0 = bitcast <4 x i32> undef to <16 x i8>
        %t12 = shufflevector <16 x i8> %0, <16 x i8> <i8 0, i8 0, i8 0, i8 0,
i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x
i32> <i32 0, i32 16, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0,
i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
        %t11 = bitcast <16 x i8> %t12 to <2 x i64>
        %t9 = extractelement <2 x i64> %t11, i32 0
        %t10 = bitcast i64 %t9 to <2 x i32>
        %t7 = bitcast i64 %test to <2 x i32>
        %t6 = xor <2 x i32> %t10, %t7
        %t1 = extractelement <2 x i32> %t6, i32 0
        ret i32 %t1
}
---------------------------------------------


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 08:41:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 08:41:08 -0500
Subject: [LLVMbugs] [Bug 4496] New: linker/lto: assertion fails when linking
 FreeBSD kernel
Message-ID: <bug-4496-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4496

           Summary: linker/lto: assertion fails when linking FreeBSD kernel
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3144)
 --> (http://llvm.org/bugs/attachment.cgi?id=3144)
tar/gz: in_pcb.o in_proto.o ip_icmp.o ip_input.o

Roman Divacky reported that when linking the FreeBSD kernel with gold using the
LLVM plugin there is an assertion failure. 

I reproduced the failure with llvm-ld:

llvm-ld: /home/edwin/llvm-svn/llvm/lib/VMCore/Constants.cpp:1320: void
llvm::ValueMap<ValType, TypeClass, ConstantClass,
HasLargeKey>::MoveConstantToNewSlot(ConstantClass*, typename
std::map<std::pair<const llvm::Type*, ValType>, llvm::Constant*,
std::less<std::pair<const llvm::Type*, ValType> >,
std::allocator<std::pair<const std::pair<const llvm::Type*, ValType>,
llvm::Constant*> > >::iterator) [with ValType = std::vector<llvm::Constant*,
std::allocator<llvm::Constant*> >, TypeClass = llvm::StructType, ConstantClass
= llvm::ConstantStruct, bool HasLargeKey = true]: Assertion `ATI !=
AbstractTypeMap.end() && "Abstract type not in AbstractTypeMap?"' failed.

Commandline used (the .o files are all LLVM bitcodes):
~/llvm-svn/llvm-obj/Release/bin/llvm-ld -v -o x.bc  in_pcb.o in_proto.o
ip_icmp.o ip_input.o

Unfortunately bugpoint is not able to reduce bugs in the linker (it crashes
itself), so all I can do is attach these .o files.

It seems that the bug reproduces only when linking all of these files at once,
linking some of them incrementally and then linking the final  ip_input.o
doesn't reproduce the bug.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 10:50:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 10:50:22 -0500
Subject: [LLVMbugs] [Bug 4497] New: dwarf exception information broken with
 -disable-fp-elim
Message-ID: <bug-4497-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4497

           Summary: dwarf exception information broken with -disable-fp-elim
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: miscompilation, regression
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: kamm-removethis at incasoftware.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3145)
 --> (http://llvm.org/bugs/attachment.cgi?id=3145)
bytecode and assembly files

LLVM change 73465 broke exception handling with debug info for LDC
(http://www.dsource.org/projects/ldc/ticket/333). 

We disable frame pointer elimination in debug builds and since the given
revision that makes _Unwind_RaiseException segfault at runtime. (I've tried
debugging the function and it seems the exception function argument is
overwritten with 0 while evaluating the cfa information.)

The D code is 
void main() { try { throw new Object; } catch(Object e) {} }
and I've attached (simplified) bytecode as well as working and broken assembly
files. The diff of the assembly files is very small and seems to contain mostly
the reordering of EH and stack adjustment information.

Unfortunately, you can't link and test these without a copy of the D exception
runtime (which I can provide, if desired). I hope the same bug can be found by
compiling equivalent C++ code with llvm-g++ and no frame pointer elimination.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 11:04:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 11:04:37 -0500
Subject: [LLVMbugs] [Bug 4495] instcombine fails
In-Reply-To: <bug-4495-206@http.llvm.org/bugs/>
Message-ID: <200907021604.n62G4b5d013151@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4495


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-02 11:04:36 ---
Fixed, thanks!
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090629/080021.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  2 17:09:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 2 Jul 2009 17:09:16 -0500
Subject: [LLVMbugs] [Bug 4499] New: clang rejects valid character constants
 beginning with three backslashes
Message-ID: <bug-4499-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4499

           Summary: clang rejects valid character constants beginning with
                    three backslashes
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:
int t1 = '\\\'';
int t2 = '\\\\';
int t3 = '\\\
';

All of these should be accepted (the first two are multi-character character
constants).

Additional testcase:
int t4 = '\\\';  // Invalid, but currently prints a random
                 // large number of diagnostics

Issue found by code inspection.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul  3 03:33:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 3 Jul 2009 03:33:55 -0500
Subject: [LLVMbugs] [Bug 4500] New: cmake: "make install" tries to set
 permissions on target directories
Message-ID: <bug-4500-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4500

           Summary: cmake: "make install" tries to set permissions on target
                    directories
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


For me "make install" when using cmake fails with:

Install the project...
-- Install configuration: ""
-- Installing: /usr/local/./include
CMake Error at cmake_install.cmake:36 (FILE):
  file Problem setting permissions on directory "/usr/local/./include"

Indeed I don't have permission to change permissions on /usr/local/./include.
And why would I need to?  There's no problem installing files inside that
directory, and setting their permissions.  I think the fact that cmake is
trying to set permissions on /usr/local/include is a bug.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul  3 04:06:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 3 Jul 2009 04:06:31 -0500
Subject: [LLVMbugs] [Bug 4501] New: iv-users asserts
Message-ID: <bug-4501-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4501

           Summary: iv-users asserts
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: opt
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rafael.espindola at gmail.com
                CC: llvmbugs at cs.uiuc.edu


"opt  -iv-users" fails on the input

------------------------------------------------
define void @test() {
entry:
        br label %bb7

bb7:
        %size.12 = phi i32 [ %t1, %bb7 ], [ undef, %entry ]
        %0 = load i16* undef, align 1
        %1 = lshr i16 %0, 8
        %2 = and i16 %1, 3
        %3 = zext i16 %2 to i32
        %4 = load i8* undef, align 1
        %5 = lshr i8 %4, 4
        %6 = and i8 %5, 1
        %7 = zext i8 %6 to i32
        %t1 = add i32 %3, %7
        br  label %bb7

bb10:
        ret void
}
--------------------------------------------------

The error is
lib/Support/APInt.cpp:492: bool llvm::APInt::ult(const llvm::APInt&) const:
Assertion `BitWidth == RHS.BitWidth && "Bit widths must be same for
comparison"' failed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul  4 09:09:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 4 Jul 2009 09:09:38 -0500
Subject: [LLVMbugs] [Bug 4453] TableGen asserting on Win32
In-Reply-To: <bug-4453-206@http.llvm.org/bugs/>
Message-ID: <200907041409.n64E9cKt022105@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4453


AlisdairM <public at alisdairm.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from AlisdairM <public at alisdairm.net>  2009-07-04 09:09:37 ---
This has been fixe independently of the bug report


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul  4 09:56:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 4 Jul 2009 09:56:44 -0500
Subject: [LLVMbugs] [Bug 4505] New: openssl-0.9.8k fails to build using
 llvm-gcc-4. 2 with arch x86_64
Message-ID: <bug-4505-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4505

           Summary: openssl-0.9.8k fails to build using llvm-gcc-4.2 with
                    arch x86_64
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
               URL: http://trac.macports.org/ticket/19446#comment:8
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ddkilzer at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


* SUMMARY
The openssl-0.9.8k package fails to build using llvm-gcc-4.2 on
SnowLeopard10A394 for arch x86_64 with the following error:

ld: in s_time.o, ObjectFileAddressSpace::mappedAddress(0xFFFFFFFFFFFFFFFC) not
in any section

* STEPS TO REPRODUCE
1. Download openssl-0.9.8k:  wget -c
http://openssl.org/source/openssl-0.9.8k.tar.gz
2. Extract the archive:  tar xzf openssl-0.9.8k.tar.gz
3. Change directories:  cd openssl-0.9.8k
4. Configure to build using llvm-gcc-4.2 and x86_64:  CC=llvm-gcc-4.2
./Configure darwin64-x86_64-cc
5. Build it:  make

* RESULTS
openssl-0.9.8k fails to build from source:

( :; LIBDEPS="${LIBDEPS:--Wl,-search_paths_first -L.. -lssl  -L.. -lcrypto }";
LDCMD="${LDCMD:-llvm-gcc-4.2}"; LDFLAGS="${LDFLAGS:--DOPENSSL_THREADS
-D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -arch x86_64 -O3 -fomit-frame-pointer
-DL_ENDIAN -DMD32_REG_T=int -Wall}"; LIBPATH=`for x in $LIBDEPS; do if echo $x
| grep '^ *-L' > /dev/null 2>&1; then echo $x | sed -e 's/^ *-L//'; fi; done |
uniq`; LIBPATH=`echo $LIBPATH | sed -e 's/ /:/g'`;
LD_LIBRARY_PATH=$LIBPATH:$LD_LIBRARY_PATH ${LDCMD} ${LDFLAGS} -o
${APPNAME:=openssl} openssl.o verify.o asn1pars.o req.o dgst.o dh.o dhparam.o
enc.o passwd.o gendh.o errstr.o ca.o pkcs7.o crl2p7.o crl.o rsa.o rsautl.o
dsa.o dsaparam.o ec.o ecparam.o x509.o genrsa.o gendsa.o s_server.o s_client.o
speed.o s_time.o apps.o s_cb.o s_socket.o app_rand.o version.o sess_id.o
ciphers.o nseq.o pkcs12.o pkcs8.o spkac.o smime.o rand.o engine.o ocsp.o
prime.o cms.o ${LIBDEPS} )
ld: in s_time.o, ObjectFileAddressSpace::mappedAddress(0xFFFFFFFFFFFFFFFC) not
in any section
collect2: ld returned 1 exit status
make[2]: *** [link_app.] Error 1
make[1]: *** [openssl] Error 2
make: *** [build_apps] Error 1

* REGRESSION
This is a regression from gcc-4.2 and x86_64 since openssl-0.9.8k builds fine
with that configuration.

* NOTES
Original bug filed on MacPorts. 
<http://trac.macports.org/ticket/19446#comment:8>


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul  4 12:25:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 4 Jul 2009 12:25:33 -0500
Subject: [LLVMbugs] [Bug 4501] scalar-evolution asserts
In-Reply-To: <bug-4501-206@http.llvm.org/bugs/>
Message-ID: <200907041725.n64HPXgn027855@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4501


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul  5 07:33:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 5 Jul 2009 07:33:18 -0500
Subject: [LLVMbugs] [Bug 4506] New: clang incorrectly searches
 lib/clang/1.0/include/ before system headers
Message-ID: <bug-4506-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4506

           Summary: clang incorrectly searches lib/clang/1.0/include/ before
                    system headers
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: OpenBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jsg at openbsd.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3154)
 --> (http://llvm.org/bugs/attachment.cgi?id=3154)
preprocessed output

clang seems to prefer $PREFIX/lib/clang/1.0/include to /usr/include 
leading to problems like clang stddef.h defining
typedef __typeof__(sizeof(int)) size_t;
where on OpenBSD i386/amd64 size_t is an unsigned long.

ie

#include <stdio.h>
#include <stddef.h>

int
main(void)
{
        return 0;
}

gives the preprocessed output shown in the attached file.
$ /usr/llvm/bin/clang -Wall -v test.c                                
clang version 1.0 (http://llvm.org/svn/llvm-project/cfe/trunk 74792)
Target: i386-unknown-openbsd4.6
Thread model: posix
 "/usr/llvm/bin/../libexec/clang-cc" -triple i386-unknown-openbsd4.6 -S
-disable-free -main-file-name test.c --relocati
on-model static --disable-fp-elim --unwind-tables=0 --mcpu=pentium4
--fmath-errno=1 -v -Wall -fdiagnostics-show-option 
-o /tmp/cc-TiXrxs.s -x c test.c
clang-cc version 1.0 based upon llvm 2.6svn hosted on i386-unknown-openbsd4.6
ignoring nonexistent directory "/System/Library/Frameworks"
ignoring nonexistent directory "/Library/Frameworks"
#include "..." search starts here:
#include <...> search starts here:
 /usr/llvm/lib/clang/1.0/include
 /usr/local/include
 /usr/include
End of search list.
In file included from test.c:2:
/usr/llvm/lib/clang/1.0/include/stddef.h:30:33: error: typedef redefinition
with different types
      ('typeof (sizeof(int))' (aka 'unsigned int') vs '__size_t' (aka 'unsigned
long'))
typedef __typeof__(sizeof(int)) size_t;
                                ^
In file included from test.c:1:
In file included from /usr/include/stdio.h:45:
/usr/include/sys/types.h:180:18: note: previous definition is here
typedef __size_t        size_t;
                        ^
2 diagnostics generated.

$ /usr/llvm/bin/llc -version
Low Level Virtual Machine (http://llvm.org/):
  llvm version 2.6svn
  DEBUG build with assertions.
  Built Jul  5 2009(13:30:51).

$ /usr/llvm/bin/clang -v
clang version 1.0 (http://llvm.org/svn/llvm-project/cfe/trunk 74792)
Target: i386-unknown-openbsd4.6
Thread model: posix


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul  5 13:50:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 5 Jul 2009 13:50:04 -0500
Subject: [LLVMbugs] [Bug 4506] clang headers incorrectly redefine size_t
In-Reply-To: <bug-4506-206@http.llvm.org/bugs/>
Message-ID: <200907051850.n65Io4DB008248@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4506


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #5 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-05 13:50:04 ---
We intentionally don't do that for stddef.h because we don't expect the system
to provide one (for example, Linux doesn't).  I fixed the size_t issue in
r74795, so it should work now.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 06:06:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 06:06:38 -0500
Subject: [LLVMbugs] [Bug 4507] New: Current (74823) LLVM fails to compile on
	DragonFly
Message-ID: <bug-4507-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4507

           Summary: Current (74823) LLVM fails to compile on DragonFly
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: swildner at erpicon.de
                CC: llvmbugs at cs.uiuc.edu


Current LLVM fails to compile on DragonFly:

----->8-----
gmake[2]: Entering directory `/home/s/tmp/compile/llvm/utils/TableGen'
llvm[2]: Linking Release executable tblgen (without symbols)
/home/s/tmp/compile/llvm/Release/lib/libLLVMSystem.a(Errno.o): In function
`llvm::sys::StrError()':
Errno.cpp:(.text+0x80): undefined reference to `llvm::sys::errno'
gmake[2]: *** [/home/s/tmp/compile/llvm/Release/bin/tblgen] Error 1
gmake[2]: Leaving directory `/home/s/tmp/compile/llvm/utils/TableGen'
gmake[1]: *** [TableGen/.makeall] Error 2
gmake[1]: Leaving directory `/home/s/tmp/compile/llvm/utils'
gmake: *** [all] Error 1
-----8<-----

DragonFly's <errno.h> defines errno thus:

----->8-----
extern __thread int     errno;

#include <sys/cdefs.h>
static __inline int *__error(void)
{
        return (&errno);
}

#define errno           (* __error())
-----8<-----

Regards,
Sascha


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 11:53:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 11:53:09 -0500
Subject: [LLVMbugs] [Bug 4507] Current (74823) LLVM fails to compile on
	DragonFly
In-Reply-To: <bug-4507-206@http.llvm.org/bugs/>
Message-ID: <200907061653.n66Gr9Ci028506@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4507


Jeffrey Yasskin <jyasskin at google.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Jeffrey Yasskin <jyasskin at google.com>  2009-07-06 11:53:02 ---
Well that was dumb on my part. I #included errno.h from inside the llvm::sys
namespace, so of course some symbols weren't defined. Sascha, please verify
that r74834 fixes this for you.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 13:30:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 13:30:08 -0500
Subject: [LLVMbugs] [Bug 4505] openssl-0.9.8k fails to build using
 llvm-gcc-4. 2 with arch x86_64
In-Reply-To: <bug-4505-206@http.llvm.org/bugs/>
Message-ID: <200907061830.n66IU8x5031747@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4505


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Evan Cheng <evan.cheng at apple.com>  2009-07-06 13:29:41 ---
This is already fixed. If you using the llvm-gcc that comes with Snow Leopard
seeds, please use the next seed that contains 2115.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 15:56:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 15:56:01 -0500
Subject: [LLVMbugs] [Bug 4508] New: llvm-config unable to figure out
 libraries on OpenSolaris
Message-ID: <bug-4508-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4508

           Summary: llvm-config unable to figure out libraries on
                    OpenSolaris
           Product: new-bugs
           Version: unspecified
          Platform: Sun
        OS/Version: Solaris
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pramach2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


I am trying to compile llvm (out of the svn repository) on a sparc machine
running OpenSolaris (which is nearly the same as Solaris10). I am having
troubles when with undefined symbols when the libraries are linking. It looks
like the llvm-config utility is unable to resolve which libraries are needed
for some of the tools, and so the symbols are unresolved.

Is there a particular part of llvm-config that needs to be modified for it to
work on Solaris, or is there a better fix for this?

Thanks,
Pradeep.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 16:49:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 16:49:10 -0500
Subject: [LLVMbugs] [Bug 4478] Assertion failed: (i == ti ||
 !mi->getOperand(i).isReg() || mi-> getOperand(i).getReg() != regA),
 function runOnMachineFunction, file TwoAddressInstructionPass.cpp, line 812.
In-Reply-To: <bug-4478-206@http.llvm.org/bugs/>
Message-ID: <200907062149.n66LnAWY005605@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4478


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Evan Cheng <evan.cheng at apple.com>  2009-07-06 16:49:09 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090706/080183.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 17:46:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 17:46:48 -0500
Subject: [LLVMbugs] [Bug 4510] New: -instcombine incorrectly eats certain
 vector operations
Message-ID: <bug-4510-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4510

           Summary: -instcombine incorrectly eats certain vector operations
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: arplynn at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3157)
 --> (http://llvm.org/bugs/attachment.cgi?id=3157)
Test case.

Running -mem2reg -instcombine on the attached testcase turns it (incorrectly)
into a simple ret void.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 21:56:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 21:56:29 -0500
Subject: [LLVMbugs] [Bug 4510] -instcombine incorrectly eats certain vector
	operations
In-Reply-To: <bug-4510-206@http.llvm.org/bugs/>
Message-ID: <200907070256.n672uTls014722@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4510


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-06 21:56:28 ---
The result of "%vecins103 = insertelement <9 x float> %tmp102, float %sub101,
i32 9" is undefined.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  6 21:58:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 6 Jul 2009 21:58:50 -0500
Subject: [LLVMbugs] [Bug 4511] New: ssi produces invalid phi node when
 adding sigma for invoke
Message-ID: <bug-4511-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4511

           Summary: ssi produces invalid phi node when adding sigma for
                    invoke
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: andrelct at dcc.ufmg.br
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3158)
 --> (http://llvm.org/bugs/attachment.cgi?id=3158)
testcase

To reproduce, you'll need a pass that runs createSSI on every non-void
instruction in the attached bugpoint-reduced testcase.

The problem is that SSI adds %SSI_phi to block %ppad81.i which references %0,
the invoke instruction whose "unwind to" address is %ppad81.i.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 05:27:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 05:27:41 -0500
Subject: [LLVMbugs] [Bug 4512] New: ARM/Linux ExecutionEngine segfault in
 Intercept. cpp StatSymbols::StatSymbols() constructor
Message-ID: <bug-4512-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4512

           Summary: ARM/Linux ExecutionEngine segfault in Intercept.cpp
                    StatSymbols::StatSymbols() constructor
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Keywords: crash-on-invalid, portability
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


This bug currently breaks all tools using the ExecutionEngine on ARM/Linux
lli, examples like Fibonacci, BrainF... , projects like OpenJDK+shark JIT etc..

xerxes at debian:/usr/src/build/nightly/llvm-74823/Release/bin$ ./lli
Segmentation fault
xerxes at debian:/usr/src/build/nightly/llvm-74823/Release/bin$ gdb ./lli
GNU gdb 6.8-debian
Copyright (C) 2008 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "arm-linux-gnueabi"...
(gdb) run
Starting program: /usr/src/build/nightly/llvm-74823/Release/bin/lli 
[Thread debugging using libthread_db enabled]
[New Thread 0x4001f620 (LWP 10219)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x4001f620 (LWP 10219)]
0x400bdaec in std::_Rb_tree_decrement () from /usr/lib/libstdc++.so.6
(gdb) bt
#0  0x400bdaec in std::_Rb_tree_decrement () from /usr/lib/libstdc++.so.6
#1  0x005741ac in std::_Rb_tree<std::string, std::pair<std::string const,
void*>, std::_Select1st<std::pair<std::string const, void*> >,
std::less<std::string>, std::allocator<std::pair<std::string const, void*> >
>::_M_insert_unique ()
#2  0x00574334 in std::_Rb_tree<std::string, std::pair<std::string const,
void*>, std::_Select1st<std::pair<std::string const, void*> >,
std::less<std::string>, std::allocator<std::pair<std::string const, void*> >
>::_M_insert_unique_ ()
#3  0x00573a80 in llvm::sys::DynamicLibrary::AddSymbol ()
#4  0x002f6a20 in global constructors keyed to Intercept.cpp ()
#5  0x0057e4dc in __libc_csu_init ()
#6  0x40214fb8 in __libc_start_main () from /lib/libc.so.6
#7  0x000c4334 in _start ()
(gdb) 

The problematic lines for ARM/Linux in Intercept.cpp are: 
    sys::DynamicLibrary::AddSymbol("stat", (void*)(intptr_t)stat);
    sys::DynamicLibrary::AddSymbol("fstat", (void*)(intptr_t)fstat);
    sys::DynamicLibrary::AddSymbol("lstat", (void*)(intptr_t)lstat);
    sys::DynamicLibrary::AddSymbol("stat64", (void*)(intptr_t)stat64);
    sys::DynamicLibrary::AddSymbol("\x1stat64", (void*)(intptr_t)stat64);
    sys::DynamicLibrary::AddSymbol("\x1open64", (void*)(intptr_t)open64);
    sys::DynamicLibrary::AddSymbol("\x1lseek64", (void*)(intptr_t)lseek64);
    sys::DynamicLibrary::AddSymbol("fstat64", (void*)(intptr_t)fstat64);
    sys::DynamicLibrary::AddSymbol("lstat64", (void*)(intptr_t)lstat64);
    sys::DynamicLibrary::AddSymbol("atexit", (void*)(intptr_t)atexit);
    sys::DynamicLibrary::AddSymbol("mknod", (void*)(intptr_t)mknod);


Do ARM/Linux need these lines or are they only needed for X86_64/Linux ?

Are we missing some include like:
#if defined(Linux)
#include <features.h> // __USE_EXTERN_INLINES?
#ifndef __USE_EXTERN_INLINES
#define __USE_EXTERN_INLINES
#endif
#endif 
so that stat can be resolved in sys/stat.h?

Someone have better ideas?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 11:27:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 11:27:16 -0500
Subject: [LLVMbugs] [Bug 4508] llvm-config unable to figure out libraries on
	OpenSolaris
In-Reply-To: <bug-4508-206@http.llvm.org/bugs/>
Message-ID: <200907071627.n67GRG2T019639@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4508


Edward O'Callaghan <eocallaghan at auroraux.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX




--- Comment #6 from Edward O'Callaghan <eocallaghan at auroraux.org>  2009-07-07 11:27:10 ---
Thanks,

Sorry to inform, this is indeed a broken GCC version, please see:
http://llvm.org/docs/GettingStarted.html#brokengcc

Your options are:

A.) File a bug with Sun to update there GCC version.

B.) You may however get a newer GCC for Solaris from the AuroraUX project.
http://junk.auroraux.org/gcc4.tar.7z

bash-3.2$ /opt/gcc4/bin/gcc --version | head -n 1
 gcc (GCC) 4.2.4

C.) Prepare your own GCC, however we have carefully prepared this and its not a
trivial matter to get a properly working modern GCC on Solaris that passes all
of its test suit.

Should you need any more help, your welcome to contact me or our mailing
list^[1] for support for this GCC binary.

Cheers,
Edward O'Callaghan.
[1] - http://lists.auroraux.org


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 13:20:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 13:20:56 -0500
Subject: [LLVMbugs] [Bug 4512] ARM/Linux ExecutionEngine segfault in
 Intercept.cpp StatSymbols: :StatSymbols() constructor
In-Reply-To: <bug-4512-206@http.llvm.org/bugs/>
Message-ID: <200907071820.n67IKupj024478@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4512


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Chris Lattner <clattner at apple.com>  2009-07-07 13:20:27 ---
System/DynamicLibrary is in sad shape, but this should fix this problem:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090706/080242.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 13:48:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 13:48:02 -0500
Subject: [LLVMbugs] [Bug 2257] Assertion in FP stackifier
In-Reply-To: <bug-2257-206@http.llvm.org/bugs/>
Message-ID: <200907071848.n67Im2XA025615@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2257


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Chris Lattner <clattner at apple.com>  2009-07-07 13:47:55 ---
Yep, probably fixed by Rafael


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 15:04:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 15:04:41 -0500
Subject: [LLVMbugs] [Bug 4513] New: Builtin::BIalloca not defined under MSVC
Message-ID: <bug-4513-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4513

           Summary: Builtin::BIalloca not defined under MSVC
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Keywords: build-problem, portability
          Severity: normal
          Priority: P2
         Component: Basic
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: abbeyj at gmail.com
                CC: llvmbugs at cs.uiuc.edu


When compiling under MSVC, Builtin::BIalloca is not declared.  This causes
CGBuiltin.cpp to fail to compile.

The problem is that the system headers have this:
#if     !__STDC__
/* Non-ANSI names for compatibility */
#define alloca  _alloca
#endif  /* __STDC__*/


When Builtin.h #includes Builtins.def it does so with BUILTIN defined as:
#define BUILTIN(ID, TYPE, ATTRS) BI##ID,


LIBBUILTIN is not defined so Builtins.def supplies a default definition:
#if defined(BUILTIN) && !defined(LIBBUILTIN)
#  define LIBBUILTIN(ID, TYPE, ATTRS, HEADER) BUILTIN(ID, TYPE, ATTRS)
#endif


Then the line
LIBBUILTIN(alloca, "v*z",         "f",     "stdlib.h")

expands to |BI_alloca,| instead of the intended |BIalloca,|

This happens because LIBBUILTIN is defined in terms of BUILTIN.  The
preprocessor rescans and expands the parameter before the concat.  If the line
was |BUILTIN(alloca, ...)| then it would work properly.

Any of the following should fix this:
1) Explicitly #undef alloca
2) Change Builtins.h to explictly #define LIBBUILTIN and remove the default
definition from Builtins.def
3) Change the default definition in Builtins.def to:
#  define LIBBUILTIN(ID, TYPE, ATTRS, HEADER) BUILTIN(##ID, TYPE, ATTRS)
but this isn't portable.  I don't know if there's a portable way to stop the
preprocessor from rescanning.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 15:18:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 15:18:25 -0500
Subject: [LLVMbugs] [Bug 3017] dedup (from the PARSEC benchmark) segfaults
 when compiled with LLVM 2.4 on Linux/x86-64
In-Reply-To: <bug-3017-206@http.llvm.org/bugs/>
Message-ID: <200907072018.n67KIPVL029402@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3017


Owen Anderson <resistor at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #3 from Owen Anderson <resistor at mac.com>  2009-07-07 15:18:20 ---
No.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 16:10:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 16:10:05 -0500
Subject: [LLVMbugs] [Bug 4514] New: clang cant compile gettext
Message-ID: <bug-4514-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4514

           Summary: clang cant compile gettext
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: rdivacky at freebsd.org
                CC: llvmbugs at cs.uiuc.edu


this works:

witten src# gcc -E -DLOCALEDIR=\"/usr/local/share/locale\" -DHAVE_CONFIG_H -I.
-I.. -I. -I. -I.. -I../intl -I./../intl -I../gnulib-lib -I./../gnulib-lib
-I/usr/local/include -DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe
-DMALLOC_PRODUCTION -march=native -fno-strict-aliasing -c gettext.c >
/tmp/gettext.c

witten src# cp /tmp/gettext.c .

witten src# clang -c gettext.c -o gettext-gettext.o                             

witten src# /bin/sh /usr/local/bin/libtool --tag=CC --mode=link clang
-DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe -DMALLOC_PRODUCTION -march=native
-fno-strict-aliasing -L/usr/local/lib -o gettext gettext-gettext.o
../gnulib-lib/libgrt.a ../intl/libintl.la -L/usr/local/lib -liconv
-R/usr/local/lib -L/usr/local/lib -liconv -R/usr/local/lib                     
                                                                               
                clang -DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe
-DMALLOC_PRODUCTION -march=native -fno-strict-aliasing -o .libs/gettext
gettext-gettext.o  -L/usr/local/lib ../gnulib-lib/libgrt.a
../intl/.libs/libintl.so /usr/local/lib/libiconv.so  -Wl,--rpath
-Wl,/usr/local/lib
clang: warning: argument unused during compilation:
'-DINSTALLDIR="/usr/local/bin"'
clang: warning: argument unused during compilation: '-O2'
clang: warning: argument unused during compilation: '-DMALLOC_PRODUCTION'
clang: warning: argument unused during compilation: '-march=native'
clang: warning: argument unused during compilation: '-fno-strict-aliasing'
creating gettext

witten src# 

ie. use CPP from gcc and then compile the resulting C code with clang and call
libtool (which uses clang for linking) to link it all

this does NOT work:

witten src# clang -E -DLOCALEDIR=\"/usr/local/share/locale\" -DHAVE_CONFIG_H
-I. -I.. -I. -I. -I.. -I../intl -I./../intl -I../gnulib-lib -I./../gnulib-lib
-I/usr/local/include -DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe
-DMALLOC_PRODUCTION -march=native -fno-strict-aliasing -c gettext.c >
/tmp/gettext.c  clang: warning: argument unused during compilation: '-c'

witten src# cp /tmp/gettext.c .

witten src# clang -c gettext.c -o gettext-gettext.o
                                                                               
witten src# /bin/sh /usr/local/bin/libtool --tag=CC --mode=link clang
-DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe -DMALLOC_PRODUCTION -march=native
-fno-strict-aliasing -L/usr/local/lib -o gettext gettext-gettext.o
../gnulib-lib/libgrt.a ../intl/libintl.la -L/usr/local/lib -liconv
-R/usr/local/lib -L/usr/local/lib -liconv -R/usr/local/lib
clang -DINSTALLDIR=\"/usr/local/bin\" -O2 -pipe -DMALLOC_PRODUCTION
-march=native -fno-strict-aliasing -o .libs/gettext gettext-gettext.o 
-L/usr/local/lib ../gnulib-lib/libgrt.a ../intl/.libs/libintl.so
/usr/local/lib/libiconv.so  -Wl,--rpath -Wl,/usr/local/lib
clang: warning: argument unused during compilation:
'-DINSTALLDIR="/usr/local/bin"'
clang: warning: argument unused during compilation: '-O2'
clang: warning: argument unused during compilation: '-DMALLOC_PRODUCTION'
clang: warning: argument unused during compilation: '-march=native'
clang: warning: argument unused during compilation: '-fno-strict-aliasing'
gettext-gettext.o(.text+0x81): In function `main':
: undefined reference to `_libintl_bindtextdomain'
gettext-gettext.o(.text+0x8e): In function `main':
: undefined reference to `_libintl_textdomain'
gettext-gettext.o(.text+0x1e3): In function `main':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x201): In function `main':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x27d): In function `main':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x2bd): In function `main':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x359): In function `main':
: undefined reference to `_libintl_bindtextdomain'
gettext-gettext.o(.text+0x36b): In function `main':
: undefined reference to `_libintl_dgettext'
gettext-gettext.o(.text+0x3d7): In function `main':
: undefined reference to `_libintl_bindtextdomain'
gettext-gettext.o(.text+0x42b): In function `main':
: undefined reference to `_libintl_dgettext'
gettext-gettext.o(.text+0x4d7): In function `usage':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x502): In function `usage':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x538): In function `usage':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x55c): In function `usage':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x580): In function `usage':
: undefined reference to `_libintl_gettext'
gettext-gettext.o(.text+0x5d9): more undefined references to `_libintl_gettext'
follow
clang: error: linker command failed with exit code 1 (use -v to see invocation)

witten src# 

ie. exactly the same steps as before except that I used clang's cpp. when I try
to use gcc to compile the clang's CPP produced C code it emits:

witten src# gcc -c gettext.c -o gettext-gettext.o      
line-map.c: file "gettext.c" left but not entered

I dont know what I should provide for you to fix this but feel free to ask I
can provide whatever you want


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 16:27:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 16:27:18 -0500
Subject: [LLVMbugs] [Bug 4515] New: certain enum constants have wrong type
 inside an enum definition
Message-ID: <bug-4515-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4515

           Summary: certain enum constants have wrong type inside an enum
                    definition
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:
enum a{a=1u,b=(a-2)/2};
int checkb[b==0?1:-1];

See also http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30260 .

Split off from bug 3173, so we don't lose track of it.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 16:54:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 16:54:47 -0500
Subject: [LLVMbugs] [Bug 4497] dwarf exception information broken with
	-disable-fp-elim
In-Reply-To: <bug-4497-206@http.llvm.org/bugs/>
Message-ID: <200907072154.n67LslqX000860@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4497


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #11 from Bill Wendling <wendling at apple.com>  2009-07-07 16:54:46 ---
I'm going to claim "FIXED" by this patch:

http://llvm.org/viewvc/llvm-project?rev=74952&view=rev

Please try this out and let me know if it fixes your problem. If it doesn't, we
can reopen this bug.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 17:17:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 17:17:44 -0500
Subject: [LLVMbugs] [Bug 4481] llvm-config with src!=obj omits
	-I<obj_dir>/include
In-Reply-To: <bug-4481-206@http.llvm.org/bugs/>
Message-ID: <200907072217.n67MHi1a001723@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4481


Jeffrey Yasskin <jyasskin at google.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Jeffrey Yasskin <jyasskin at google.com>  2009-07-07 17:17:43 ---
Fixed by r74956.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  7 18:52:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 7 Jul 2009 18:52:10 -0500
Subject: [LLVMbugs] [Bug 4497] dwarf exception information broken with
	-disable-fp-elim
In-Reply-To: <bug-4497-206@http.llvm.org/bugs/>
Message-ID: <200907072352.n67NqA7j004995@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4497


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #12 from Bill Wendling <wendling at apple.com>  2009-07-07 18:51:57 ---
This caused a bootstrap failure in llvm-gcc. Reopening.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 00:24:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 00:24:56 -0500
Subject: [LLVMbugs] [Bug 4517] New: missed constant fold
Message-ID: <bug-4517-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4517

           Summary: missed constant fold
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


I saw this constant expression in real code after llvm-g++ -O2:

declare extern_weak i32 @0(i64)

define void @foo() {
  br i1 icmp eq (i32 zext (i1 icmp ne (i32 (i64)* @0, i32 (i64)* null) to i32),
i32 0), label %cond_true, label %cond_false
cond_true:
  ret void
cond_false:
  ret void
}

That branch expression should be reduced to:

  i1 icmp eq (i32 (i64)* @0, i32 (i64)* null)

It's probably not a perf issue, I just happened to see it while examining
something else and didn't want to forget about it.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 00:50:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 00:50:40 -0500
Subject: [LLVMbugs] [Bug 4518] New: weakref emits undef symbol
Message-ID: <bug-4518-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4518

           Summary: weakref emits undef symbol
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu, nlewycky at google.com


Consider this program short C program:

  int target_func(unsigned long);
  static int bad_func(unsigned long)
__attribute__((__weakref__("target_func")));
  extern void foo(int);

  void user() {
    if (bad_func) {
      foo(0);
    } else {
      foo(1);
    }
  }

This emits to LLVM @bad_func = alias weak ... @target_func and the @user
function references @bad_func. Instead the frontend should turn references to
@bad_func into references to @target_func and never emit the name 'bad_func' to
the LLVM IR.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 01:20:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 01:20:21 -0500
Subject: [LLVMbugs] [Bug 4517] missed constant fold
In-Reply-To: <bug-4517-206@http.llvm.org/bugs/>
Message-ID: <200907080620.n686KLrC016798@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4517


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |LATER




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-08 01:20:20 ---
Please move this to a readme.txt file.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 09:07:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 09:07:36 -0500
Subject: [LLVMbugs] [Bug 4514] clang cant compile gettext
In-Reply-To: <bug-4514-206@http.llvm.org/bugs/>
Message-ID: <200907081407.n68E7ax6011507@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4514


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2009-07-08 09:05:58 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090706/018832.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 11:28:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 11:28:11 -0500
Subject: [LLVMbugs] [Bug 4513] Builtin::BIalloca not defined under MSVC
In-Reply-To: <bug-4513-206@http.llvm.org/bugs/>
Message-ID: <200907081628.n68GSBOP016088@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4513


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Chris Lattner <clattner at apple.com>  2009-07-08 11:26:24 ---
Looks great, applied, thanks!
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090706/018833.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 14:45:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 14:45:06 -0500
Subject: [LLVMbugs] [Bug 4521] New: Internal compiler error in
	SelectionDAGBuild.cpp
Message-ID: <bug-4521-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4521

           Summary: Internal compiler error in SelectionDAGBuild.cpp
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Archive library
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: wjjeon at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3163)
 --> (http://llvm.org/bugs/attachment.cgi?id=3163)
source code in Linux OMAP kernel

Hello,

While I was trying to cross-compile Linux OMAP kernel with llvm, I have the
following error message.

  CC      arch/arm/kernel/traps.o

  cc1: /home/wonjeon/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp:5388:
void llvm::SelectionDAGLowering::visitInlineAsm(llvm::CallSite): Assertion
`(OpInfo.ConstraintType == TargetLowering::C_RegisterClass ||
OpInfo.ConstraintType == TargetLowering::C_Register) && "Unknown constraint
type!"' failed.

  arch/arm/kernel/traps.c:748: internal compiler error: Aborted

  Please submit a full bug report,

  with preprocessed source if appropriate.

  See <URL:http://llvm.org/bugs/> for instructions



Is there any related bug filed or fixed? traps.c has been attached. Thanks in
advance.



Won


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 16:05:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 16:05:11 -0500
Subject: [LLVMbugs] [Bug 4497] dwarf exception information broken with
	-disable-fp-elim
In-Reply-To: <bug-4497-206@http.llvm.org/bugs/>
Message-ID: <200907082105.n68L5BN4026892@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4497


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #13 from Bill Wendling <wendling at apple.com>  2009-07-08 16:05:04 ---
Fixed:

http://llvm.org/viewvc/llvm-project?rev=75047&view=rev

Please verify! :-)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 16:07:51 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 16:07:51 -0500
Subject: [LLVMbugs] [Bug 4522] New: cmake builds uses configure builds
 tblgen output instead of generating its own
Message-ID: <bug-4522-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4522

           Summary: cmake builds uses configure builds tblgen output instead
                    of generating its own
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: build-problem
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


quite often when someone have updated any tablegen files i face cmake errors
like these:

Scanning dependencies of target ARMCodeGenTable_gen
[ 42%] Building ARMGenRegisterInfo.h.inc.tmp...
[ 42%] Building ARMGenRegisterInfo.h.inc...
[ 42%] Building ARMGenRegisterNames.inc.tmp...
[ 42%] Building ARMGenRegisterNames.inc...
[ 42%] Building ARMGenRegisterInfo.inc.tmp...
[ 42%] Building ARMGenRegisterInfo.inc...
[ 42%] Building ARMGenInstrNames.inc.tmp...
[ 42%] Building ARMGenInstrNames.inc...
[ 42%] Building ARMGenInstrInfo.inc.tmp...
[ 42%] Building ARMGenInstrInfo.inc...
[ 42%] Building ARMGenCodeEmitter.inc.tmp...
[ 42%] Building ARMGenCodeEmitter.inc...
[ 42%] Building ARMGenAsmWriter.inc.tmp...
[ 42%] Building ARMGenAsmWriter.inc...
[ 42%] Building ARMGenDAGISel.inc.tmp...
[ 42%] Building ARMGenDAGISel.inc...
[ 42%] Building ARMGenCallingConv.inc.tmp...
[ 42%] Building ARMGenCallingConv.inc...
[ 42%] Building ARMGenSubtarget.inc.tmp...
[ 42%] Building ARMGenSubtarget.inc...
[ 44%] Built target ARMCodeGenTable_gen
Scanning dependencies of target LLVMARMCodeGen
[ 44%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMBaseInstrInfo.cpp.o
[ 44%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMCodeEmitter.cpp.o
[ 44%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMConstantIslandPass.cpp.o
[ 45%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMConstantPoolValue.cpp.o
[ 45%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMInstrInfo.cpp.o
[ 45%] Building CXX object
lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMISelDAGToDAG.cpp.o
/home/xranby/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp: In member function
'llvm::SDNode*<unnamed>::ARMDAGToDAGISel::Select(llvm::SDValue)':
/home/xranby/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1057: error: 't2MOVCCs' is
not a member of 'llvm::ARM'
/home/xranby/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1088: error: 't2MOVCCi' is
not a member of 'llvm::ARM'
/home/xranby/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp:1126: error: 't2MOVCCr' is
not a member of 'llvm::ARM'
make[2]: ***
[lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/ARMISelDAGToDAG.cpp.o] Error 1
make[1]: *** [lib/Target/ARM/CMakeFiles/LLVMARMCodeGen.dir/all] Error 2

Instinctively when i see these cmake errors i try a configure build in the
source directory to check if anything works.. the configure builds then
succeed.. so ok back to cmake again to check what wnet wrong in the first place

And now my cmake build succeeded as well without any sourcecode changes.

The only conclusion i can make of this are that the cmake build uses some
output from the configure build
my guess are that it uses the configure builds tablegen generated inc files.

Could it be that my cmake build gets confused if i have previously run a
configure build in the llvm sourcetree that cmake fetches its sources from?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul  8 22:12:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 8 Jul 2009 22:12:20 -0500
Subject: [LLVMbugs] [Bug 4523] New: llvm-gcc generates symbol for weakref
Message-ID: <bug-4523-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4523

           Summary: llvm-gcc generates symbol for weakref
           Product: tools
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


This code shouldn't generate an alias on Darwin (note that it should on Linux).

  int target_func(unsigned long);
  static int bad_func(unsigned long)
__attribute__((__weakref__("target_func")));
  extern void foo(int);

  void user() {
    if (bad_func) {
      foo(0);
    } else {
      foo(1);
    }
  }

which generates:

  declare extern_weak i32 @bad_func(i32)

  define void @user() nounwind {
  entry:
    br i1 icmp ne (i32 (i32)* @bad_func, i32 (i32)* null), label %bb, label
%bb1
  [...]

it shouldn't mention bad_func at all. If you patch this, please be sure not to
change the behaviour on Linux...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 01:15:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 01:15:54 -0500
Subject: [LLVMbugs] [Bug 4518] libLTO emits undefined symbol for weakref
In-Reply-To: <bug-4518-206@http.llvm.org/bugs/>
Message-ID: <200907090615.n696Fshr012874@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4518


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 12:30:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 12:30:50 -0500
Subject: [LLVMbugs] [Bug 4497] dwarf exception information broken with
	-disable-fp-elim
In-Reply-To: <bug-4497-206@http.llvm.org/bugs/>
Message-ID: <200907091730.n69HUoC8014148@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4497


Christian Kamm <kamm-removethis at incasoftware.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #14 from Christian Kamm <kamm-removethis at incasoftware.de>  2009-07-09 12:30:50 ---
Now the code works correctly with -disable-fp-elim.

However, it fails without it now: when I compile the exception handling runtime
(that contains a function calling _Unwind_RaiseException) without
-disable-fp-elim, _Unwind_RaiseException returns an error code of 5 (which I
think means END_OF_STACK). 

It does not matter whether the user code is compiled with -disable-fp-elim or
not. This case worked correctly prior to the fix.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 15:26:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 15:26:33 -0500
Subject: [LLVMbugs] [Bug 4524] New: Cmake builds do not include the MSP430.
Message-ID: <bug-4524-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4524

           Summary: Cmake builds do not include the MSP430.
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


The enclosed patch adds the MSP430 to the main CMakeLists.txt file and
removes ab extraneous file from the MSP430's CMakeLists.txt.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 16:04:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 16:04:23 -0500
Subject: [LLVMbugs] [Bug 4525] New: llvm-gcc does not compile
Message-ID: <bug-4525-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4525

           Summary: llvm-gcc does not compile
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pramach2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


Hi,
I am trying to compile a version of llvm-gcc that I just checked out of the svn
repository, with --enable-llvm pointing to the latest llvm version checked out
from the repository, and I am having problems. I get the following error

c++ -c -g -DIN_GCC -W -Wall -Wwrite-strings -pedantic -Wno-long-long
-Wno-variadic-macros -Wmissing-format-attribute -fno-common -DHAVE_CONFIG_H
-Wno-unused -DTARGET_NAME=\"sparc-sun-solaris2.11\" -I. -I.
-I../../llvm-gcc-4.2/gcc -I../../llvm-gcc-4.2/gcc/.
-I../../llvm-gcc-4.2/gcc/../include -I../../llvm-gcc-4.2/gcc/../libcpp/include 
-I../../llvm-gcc-4.2/gcc/../libdecnumber -I../libdecnumber
-I/home/pramach2/research/llvm-solaris/obj/include
-I/home/pramach2/research/llvm-solaris/src/include -DENABLE_LLVM
-I/home/pramach2/research/llvm-solaris/obj/../src/include  -D_DEBUG  -include
llvm/System/Solaris.h -D_GNU_SOURCE -D__STDC_LIMIT_MACROS
-D__STDC_CONSTANT_MACROS   -I. -I. -I../../llvm-gcc-4.2/gcc
-I../../llvm-gcc-4.2/gcc/. -I../../llvm-gcc-4.2/gcc/../include
-I../../llvm-gcc-4.2/gcc/../libcpp/include 
-I../../llvm-gcc-4.2/gcc/../libdecnumber -I../libdecnumber
-I/home/pramach2/research/llvm-solaris/obj/include
-I/home/pramach2/research/llvm-solaris/src/include
../../llvm-gcc-4.2/gcc/llvm-backend.cpp -o llvm-backend.o
../../llvm-gcc-4.2/gcc/llvm-backend.cpp: In function ???????void
writeLLVMValues()???????:
../../llvm-gcc-4.2/gcc/llvm-backend.cpp:292: error: no matching function for
call to ???????llvm::GlobalVariable::GlobalVariable(llvm::Module&, const
llvm::Type*, bool, llvm::GlobalValue::LinkageTypes, llvm::Constant*&, const
char [16])???????
/home/pramach2/research/llvm-solaris/src/include/llvm/GlobalVariable.h:61:
note: candidates are: llvm::GlobalVariable::GlobalVariable(const llvm::Type*,
bool, llvm::GlobalValue::LinkageTypes, llvm::Constant*, const std::string&,
llvm::GlobalVariable*, bool, unsigned int)
/home/pramach2/research/llvm-solaris/src/include/llvm/GlobalVariable.h:55:
note:                 llvm::GlobalVariable::GlobalVariable(const llvm::Type*,
bool, llvm::GlobalValue::LinkageTypes, llvm::Constant*, const std::string&,
llvm::Module*, bool, unsigned int)
/home/pramach2/research/llvm-solaris/src/include/llvm/GlobalVariable.h:38:
note:                 llvm::GlobalVariable::GlobalVariable(const
llvm::GlobalVariable&)

I have gcc version 4.1.2 on a i386 machine running Linux. I have configured
llvm for Debug build, with the --enable-checking flag.

I have also tried doing the same under sparc/Solaris (I have gcc 4.2.2 and the
latest version of llvm compiled on Sparc/solaris) and I run in to the same
problem, but the following error also occurs in Sparc.

../../llvm-gcc-4.2/gcc/llvm-backend.cpp:333:2: error: #error LLVM_TARGET_NAME
macro not specified by GCC backend

Could someone help me sort this problem out?

Thanks,
pradeep.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 17:01:24 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 17:01:24 -0500
Subject: [LLVMbugs] [Bug 3037] LegalizeTypes fails on i1 = extractelement <n
	x i1>, N
In-Reply-To: <bug-3037-206@http.llvm.org/bugs/>
Message-ID: <200907092201.n69M1Oo5024137@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3037


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-09 17:01:23 ---
Fixed in r75176.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 17:33:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 17:33:25 -0500
Subject: [LLVMbugs] [Bug 4497] dwarf exception information broken with
	-disable-fp-elim
In-Reply-To: <bug-4497-206@http.llvm.org/bugs/>
Message-ID: <200907092233.n69MXPc0025219@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4497


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #18 from Bill Wendling <wendling at apple.com>  2009-07-09 17:33:25 ---
Okay. Should be fixed now:

http://llvm.org/viewvc/llvm-project?rev=75183&view=rev


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 19:11:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 19:11:25 -0500
Subject: [LLVMbugs] [Bug 4521] inline asm "p" constraint not handled
In-Reply-To: <bug-4521-206@http.llvm.org/bugs/>
Message-ID: <200907100011.n6A0BPFJ028718@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4521


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Bob Wilson <bob.wilson at apple.com>  2009-07-09 19:11:16 ---
This should be fixed now in:
http://llvm.org/viewvc/llvm-project?rev=75201&view=rev
http://llvm.org/viewvc/llvm-project?rev=75203&view=rev


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  9 21:31:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 9 Jul 2009 21:31:34 -0500
Subject: [LLVMbugs] [Bug 4524] Cmake builds do not include the MSP430.
In-Reply-To: <bug-4524-206@http.llvm.org/bugs/>
Message-ID: <200907100231.n6A2VYcJ000987@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4524


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-07-09 21:31:31 ---
I think this got applied


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 01:58:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 01:58:19 -0500
Subject: [LLVMbugs] [Bug 4526] New: clang: should support macro definitions
 when -g3 is used
Message-ID: <bug-4526-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4526

           Summary: clang: should support macro definitions when -g3 is used
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


#define X(a) (a+1)
int main()
{
    return X(1);
}

Compiled with gcc -g3 I am able to evaluate the X() macro in gdb:

$ gcc -g3 x.c
$ gdb ./a.out
...
(gdb) b main
(gdb) r
(gdb) p X(1)
$1 = 2
(gdb)

Compiled with clang -g3, it behaves the same as -g: I don't get macro
information:
$ clang -g3 x.c
$ gdb ./a.out
...
(gdb) b main
(gdb) r
(gdb) p X(1)
No symbol "X" in current context.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 10:39:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 10:39:40 -0500
Subject: [LLVMbugs] [Bug 4528] New: Internal compiler error in
	RegisterScavenging.cpp
Message-ID: <bug-4528-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4528

           Summary: Internal compiler error in RegisterScavenging.cpp
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Archive library
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: wjjeon at gmail.com
                CC: llvmbugs at cs.uiuc.edu


While I was trying to compile the source code, I have the following error
message. All the compiler options look like this:

  llvm-arm-gcc -Wp,-MD,mm/.filemap.o.d  -nostdinc -isystem
/home/wonjeon/llvm-gcc-arm-install/lib/gcc/arm-none-linux-gnueabi/4.2.1/include
-Iinclude  -I/home/wonjeon/linux-omap-2.6/arch/arm/include -include
include/linux/autoconf.h -D__KERNEL__ -mlittle-endian
-Iarch/arm/mach-omap2/include -Iarch/arm/plat-omap/include -Wall -Wundef
-Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Os -marm -mabi=aapcs-linux
-mno-thumb-interwork -funwind-tables -D__LINUX_ARM_ARCH__=6 -march=armv6
-mtune=arm1136j-s -msoft-float -Uarm -fno-stack-protector -fomit-frame-pointer
-Wdeclaration-after-statement -Wno-pointer-sign -fwrapv   -D"KBUILD_STR(s)=#s"
-D"KBUILD_BASENAME=KBUILD_STR(filemap)"  -D"KBUILD_MODNAME=KBUILD_STR(filemap)"
 -c -o mm/filemap.o mm/filemap.c
cc1: /home/wonjeon/llvm-src/lib/CodeGen/RegisterScavenging.cpp:258: void
llvm::RegScavenger::forward(): Assertion `isUsed(Reg) && "Using an undefined
register!"' failed.
mm/filemap.c:2530: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

Can anyone take a look at it? filemap.i has been attached. If you need more
information, please let me know.

Thanks,
Won


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 13:22:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 13:22:01 -0500
Subject: [LLVMbugs] [Bug 4525] llvm-gcc does not compile
In-Reply-To: <bug-4525-206@http.llvm.org/bugs/>
Message-ID: <200907101822.n6AIM1iF011657@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4525


Pradeep Ramachandran <pramach2 at uiuc.edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX




--- Comment #3 from Pradeep Ramachandran <pramach2 at uiuc.edu>  2009-07-10 13:21:57 ---
I just realized that I was using a broken version of gcc (4.2.1) on my X86_64
machine. The docs say that this is unsupported. I have changed the bug status
to "wontfix"


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 13:26:30 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 13:26:30 -0500
Subject: [LLVMbugs] [Bug 4525] llvm-gcc does not compile
In-Reply-To: <bug-4525-206@http.llvm.org/bugs/>
Message-ID: <200907101826.n6AIQUb6011895@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4525


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WONTFIX                     |




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2009-07-10 13:26:29 ---
I can reproduce this if I build llvm with the right options.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 13:43:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 13:43:28 -0500
Subject: [LLVMbugs] [Bug 4529] New: llvm-gcc 4.2 on Solaris gives internal
 compiler error
Message-ID: <bug-4529-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4529

           Summary: llvm-gcc 4.2 on Solaris gives internal compiler error
           Product: new-bugs
           Version: unspecified
          Platform: Sun
        OS/Version: Solaris
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pramach2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


I am trying to compile llvm-gcc 4.2-2.5 source on sparc-sun-solaris11 (it is a
Niagara T2 running OpenSolais, which is practically the same as Solaris 10).
Release build for llvm 2.5, however, built with no problems. I am using gcc
4.2.2 on the solaris machine (built from source) and have configured llvm-gcc
with

--prefix=`pwd`/../install --program-prefix=llvm
--enable-llvm=`pwd`/../../llvm/obj --enable-languages=c --disable-shared
--disable-multilib --disable-bootstrap --with-as=/usr/sfw/bin/gas

This is the same bug that was previously reported on llvm-devel for
sparc-sun-solaris10
<http://lists.cs.uiuc.edu/pipermail/llvmdev/2009-March/020794.html>. A snippet
is pasted below.

Does anyone know of a fix for this bug?

Thanks,
Pradeep.

gmake[4]: Entering directory
`/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/gcc'
gmake[4]: Warning: File `crtbegin.o' has modification time 77 s in the future
gmake[4]: `crtend.o' is up to date.
gmake[4]: `gmon.o' is up to date.
gmake[4]: `crt1.o' is up to date.
gmake[4]: `crti.o' is up to date.
gmake[4]: `crtn.o' is up to date.
gmake[4]: `gcrt1.o' is up to date.
gmake[4]: `crtfastmath.o' is up to date.
gmake[4]: warning:  Clock skew detected.  Your build may be incomplete.
gmake[4]: Leaving directory
`/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/gcc'
/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/./gcc/xgcc
-B/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/./gcc/
-B/llvm-gcc-install/sparc-sun-solaris2.11/bin/
-B/llvm-gcc-install/sparc-sun-solaris2.11/lib/ -isystem
/llvm-gcc-install/sparc-sun-solaris2.11/include -isystem
/llvm-gcc-install/sparc-sun-solaris2.11/sys-include  -O2  -O2 -g -O2  -DIN_GCC 
  -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition  -isystem ./include  -fPIC -g -DHAVE_GTHR_DEFAULT
-DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  -I. -I.
-I../../llvm-gcc4.2-2.5.source/gcc -I../../llvm-gcc4.2-2.5.source/gcc/.
-I../../llvm-gcc4.2-2.5.source/gcc/../include
-I../../llvm-gcc4.2-2.5.source/gcc/../libcpp/include 
-I../../llvm-gcc4.2-2.5.source/gcc/../libdecnumber -I../libdecnumber
-I/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-build/include
-I/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-2.5/include
-DL_powitf2 -c ../../llvm-gcc4.2-2.5.source/gcc/libgcc2.c -o
libgcc/./_powitf2.o
../../llvm-gcc4.2-2.5.source/gcc/libgcc2.c: In function
???????__powitf2???????:
../../llvm-gcc4.2-2.5.source/gcc/libgcc2.c:1765: internal compiler error: in
HandleArgument, at llvm-abi.h:520
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.
gmake[3]: *** [libgcc/./_powitf2.o] Error 1
gmake[3]: Leaving directory
`/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/gcc'
gmake[2]: *** [stmp-multilib] Error 2
gmake[2]: Leaving directory
`/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build/gcc'
gmake[1]: *** [all-gcc] Error 2
gmake[1]: Leaving directory
`/home/pramach2/research/SoftBound-v1.1-solaris/SoftBound-v1.1/llvm-gcc-build'
gmake: *** [all] Error 2


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 14:50:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 14:50:07 -0500
Subject: [LLVMbugs] [Bug 4530] New: llvm-ld does not merge debug information
	correctly
Message-ID: <bug-4530-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4530

           Summary: llvm-ld does not merge debug information correctly
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-ld
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: haohui.mai at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3169)
 --> (http://llvm.org/bugs/attachment.cgi?id=3169)
Test case

Debug information are messed up after llvm-ld. There are a number of problems:

1. There is only one compilation unit in the linked bitcode file, and all debug
information points to this compilation unit, which gives incorrect results on
source filenames.

2. The line number are messed up too. I didn't dig into it.

The result is still incorrect even with all optimizations disabled.

I think simply marking compilation unit as internal globals instead of link
once would solve problem No.1, if the code generator does not recognize
multiple main compilation unit, just clean up the flag would work.

However, I have no idea on how to deal with problem No.2.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 15:24:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 15:24:06 -0500
Subject: [LLVMbugs] [Bug 4389] clang: wrong scoping: refuses to compile
 global variable name same as local variable name
In-Reply-To: <bug-4389-206@http.llvm.org/bugs/>
Message-ID: <200907102024.n6AKO6ao016486@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4389


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-10 15:24:05 ---
This was fixed in r74917.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 19:14:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 19:14:02 -0500
Subject: [LLVMbugs] [Bug 4532] New: llc infloops in
	FoldingSetImpl::RemoveNode in llc
Message-ID: <bug-4532-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4532

           Summary: llc infloops in FoldingSetImpl::RemoveNode in llc
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


This shows up during "make check" on ARM.

llvm-as < test/CodeGen/2006-11-10-CycleInDAG.ll | llc -march=arm -mattr=+v6
[... wait forever ...]

0x007e5968 in llvm::FoldingSetImpl::RemoveNode ()
(gdb) bt
#0  0x007e5968 in llvm::FoldingSetImpl::RemoveNode ()
#1  0x004dce50 in llvm::SelectionDAG::RemoveNodeFromCSEMaps ()
#2  0x004ddd10 in llvm::SelectionDAG::DeleteNode ()
#3  0x004ba1dc in llvm::SelectionDAG::Combine ()
#4  0x00521d50 in llvm::SelectionDAGISel::CodeGenAndEmitDAG ()
#5  0x0052314c in llvm::SelectionDAGISel::SelectBasicBlock ()
#6  0x00523f78 in llvm::SelectionDAGISel::SelectAllBasicBlocks ()
#7  0x00525224 in llvm::SelectionDAGISel::runOnFunction ()
#8  0x00796f48 in llvm::FPPassManager::runOnFunction ()
#9  0x007971f0 in llvm::FunctionPassManagerImpl::run ()
#10 0x007973f8 in llvm::FunctionPassManager::run ()
#11 0x000d5a04 in main ()

The call to RemoveNode never finishes. This is with a Release-Asserts build,
built with "gcc version 4.3.3 (Debian 4.3.3-10)".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 10 21:02:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 10 Jul 2009 21:02:52 -0500
Subject: [LLVMbugs] [Bug 4533] New: relocation problem with shared libraries
	on x86-64
Message-ID: <bug-4533-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4533

           Summary: relocation problem with shared libraries on x86-64
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu


The problem is that the address of a function shows up differently in two
different translation units. One TU is compiled to a shared library with -fPIC,
the other has neither. This works with GCC but not LLVM:

$ g++ -O2 -fPIC x.cc -shared -o x.so
$ g++ -O2 y.cc -o y x.so
$ ./y
y: 0x400570
x: 0x400570

$ llvm-g++ -O2 -fPIC x.cc -shared -o x.so
$ llvm-g++ -O2 y.cc -o y x.so
$ ./y
y: 0x4004c8
x: 0x7f07903b2240

The testcase is two very small files:

x.cc:
  #include <cstdio>

  void trouble() {}

  void func() {
    printf("x: %p\n", trouble);
  }

y.cc:
  #include <cstdio>

  void trouble();
  extern void func();

  int main(void) {
    printf("y: %p\n", trouble);
    func();
  }


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 00:34:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 00:34:48 -0500
Subject: [LLVMbugs] [Bug 4534] New: LLVM taking too long to build a file
Message-ID: <bug-4534-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4534

           Summary: LLVM taking too long to build a file
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: critical
          Priority: P2
         Component: Loop Optimizer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: revcompgeek at gmail.com
                CC: llvmbugs at cs.uiuc.edu, revcompgeek at gmail.com


Simply running llc on this file seems to take way too long, and doesn't finish
within an hour. A sample backtrace from llc that was taken by gdb while it was
running as well as a backtrace from LDC are attached.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 02:29:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 02:29:20 -0500
Subject: [LLVMbugs] [Bug 4535] New: mark constant strings private instead of
	internal
Message-ID: <bug-4535-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4535

           Summary: mark constant strings private instead of internal
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu


We should use the private linkage for constant strings instead of internal to
avoid bloated .o files.

Chris says, this is the ifdefs in TreeConstantToLLVM::EmitLV_STRING_CST.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 06:21:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 06:21:50 -0500
Subject: [LLVMbugs] [Bug 4536] New: clang __gnu_inline__ semantics do not
	match gcc
Message-ID: <bug-4536-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4536

           Summary: clang __gnu_inline__ semantics do not match gcc
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: OpenBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jsg at openbsd.org
                CC: llvmbugs at cs.uiuc.edu


int foo(void);

extern inline __attribute__((__gnu_inline__)) int
foo(void)
{
        return 5;
}

int
main(int argc, char *argv[])
{
        foo();
        return (0);
}

will cause a symbol to be generated with clang if the prototype aka "int
foo(void);" is present.

ie
# nm foo.o
00000000 T foo
00000000 F foo.c
00000020 T main

if the prototype isn't present then a symbol isn't generated.
# nm foo.o 
         U foo
00000000 F foo.c
00000000 T main

gcc 4.2.4 will apparently never create a symbol in both cases:
# nm foo.o 
         U foo
00000000 F foo.c
00000000 T main


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 11:37:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 11:37:19 -0500
Subject: [LLVMbugs] [Bug 4537] New: infinite loop in scev
Message-ID: <bug-4537-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4537

           Summary: infinite loop in scev
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Loop Optimizer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: gohman at apple.com, llvmbugs at cs.uiuc.edu


Created an attachment (id=3178)
 --> (http://llvm.org/bugs/attachment.cgi?id=3178)
testcase

$ opt -analyze -scalar-evolution b.bc -disable-output
Printing analysis 'Scalar Evolution Analysis' for function 'test':
Segmentation fault

I'll skip the nonsense and get straight to the infloopy part of the backtrace:

#19 0x082caa85 in llvm::ScalarEvolution::getSignExtendExpr (this=0x883e938, 
    Op=0x8841310, Ty=0x883b280) at ScalarEvolution.cpp:1000
#20 0x082c9070 in llvm::ScalarEvolution::createSCEV (this=0x883e938, 
    V=0x883fcec) at ScalarEvolution.cpp:2928
#21 0x082c9946 in llvm::ScalarEvolution::getSCEV (this=0x883e938, V=0x883fcec)
    at ScalarEvolution.cpp:2130
#22 0x082c807a in llvm::ScalarEvolution::createSCEV (this=0x883e938, 
    V=0x883fd30) at ScalarEvolution.cpp:2788
#23 0x082c9946 in llvm::ScalarEvolution::getSCEV (this=0x883e938, V=0x883fd30)
    at ScalarEvolution.cpp:2130
#24 0x082c9ea4 in llvm::ScalarEvolution::isNecessaryCond (this=0x883e938, 
    CondValue=0x8840008, Pred=llvm::CmpInst::ICMP_SGT, LHS=0x8841310, 
    RHS=0x8841594, Inverse=true) at ScalarEvolution.cpp:4530
#25 0x082c9b6f in llvm::ScalarEvolution::isNecessaryCond (this=0x883e938, 
    CondValue=0x8840098, Pred=llvm::CmpInst::ICMP_SGT, LHS=0x8841310, 
    RHS=0x8841594, Inverse=true) at ScalarEvolution.cpp:4449
#26 0x082ca327 in llvm::ScalarEvolution::isLoopBackedgeGuardedByCond (
    this=0x883e938, L=0x883df90, Pred=llvm::CmpInst::ICMP_SGT, LHS=0x8841310, 
    RHS=0x8841594) at ScalarEvolution.cpp:4395
#27 0x082caa85 in llvm::ScalarEvolution::getSignExtendExpr (this=0x883e938, 
    Op=0x8841310, Ty=0x883b280) at ScalarEvolution.cpp:1000

At #19, MaxBECount is i32 2147483646.
At #20, V and U are '%6 = sext i32 %5 to i64'.
At #23, V is %7 = udiv i64 undef, %6
At #24, PreCondLHS = %7, PreCondRHS = i64 0, Pred = ICMP_SGT.
At #25, BO is '%or.cond.i = or i1 %phitmp, %.not.i', LHS = '{(trunc i64 ((-1 *
%0) /u 8) to i32),+,-1}<bb4.i>', RHS = '-2147483648', Pred = ICMP_SGT and
Inverse = true.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 14:23:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 14:23:09 -0500
Subject: [LLVMbugs] [Bug 4538] New: opt -iv-users dies with Illegal
	instruction
Message-ID: <bug-4538-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4538

           Summary: opt -iv-users dies with Illegal instruction
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Created an attachment (id=3179)
 --> (http://llvm.org/bugs/attachment.cgi?id=3179)
bugpoint-reduced-simplified.bc

# opt bugpoint-reduced-simplified.bc -iv-users
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

Illegal instruction (core dumped)

Found when building vfs_subr.c from the FreeBSD kernel.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 14:33:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 14:33:40 -0500
Subject: [LLVMbugs] [Bug 4538] opt -iv-users dies with Illegal instruction
In-Reply-To: <bug-4538-206@http.llvm.org/bugs/>
Message-ID: <200907111933.n6BJXebQ015577@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4538


Pawel Worach <pawel.worach at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Pawel Worach <pawel.worach at gmail.com>  2009-07-11 14:33:39 ---


*** This bug has been marked as a duplicate of bug 4537 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 15:29:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 15:29:53 -0500
Subject: [LLVMbugs] [Bug 4533] relocation problem with shared libraries on
	x86-64
In-Reply-To: <bug-4533-206@http.llvm.org/bugs/>
Message-ID: <200907112029.n6BKTrBP017730@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4533


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Chris Lattner <clattner at apple.com>  2009-07-11 15:29:52 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090706/080739.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 15:35:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 15:35:26 -0500
Subject: [LLVMbugs] [Bug 4539] New: Verifier: check for unsized TLS objects
Message-ID: <bug-4539-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4539

           Summary: Verifier: check for unsized TLS objects
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: clattner at apple.com, llvmbugs at cs.uiuc.edu


On 2009-07-08 23:00, Chris Lattner wrote:
>> +++ llvm/trunk/lib/Target/XCore/XCoreISelLowering.cpp Wed Jul  8  
>> 14:04:27 2009
>> @@ -270,9 +270,8 @@
>>   }
>>   const Type *Ty = cast<PointerType>(GV->getType())->getElementType();
>>   if (!Ty->isSized() || isZeroLengthArray(Ty)) {
>> -    cerr << "Size of thread local object " << GVar->getName()
>> -         << " is unknown\n";
>> -    abort();
>> +    llvm_report_error("Size of thread local object " + GVar- 
>>> getName()
>> +                      + " is unknown");
>
> This should be an assertion, does the verifier catch this?  If not, it  
> should.

I don't see any code in the verifier to check Thread Local variables.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 19:58:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 19:58:54 -0500
Subject: [LLVMbugs] [Bug 4540] New: Change in case statement value generates
 incorrect code ( crashes the runtime linker)
Message-ID: <bug-4540-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4540

           Summary: Change in case statement value generates incorrect code
                    (crashes the runtime linker)
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: major
          Priority: P2
         Component: Backend: PowerPC
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: revcompgeek at gmail.com
                CC: llvmbugs at cs.uiuc.edu, revcompgeek at gmail.com


Created an attachment (id=3181)
 --> (http://llvm.org/bugs/attachment.cgi?id=3181)
Archive of the related files.

GDB reports that dyld, the dynamic runtime linker crashes when running an
executable generated from the switch_bad.ll file in the attachment. Line 109 is
the only thing that is changed between the good.ll file and the bad.ll file,
and the only change is in the value it is testing for. Some values for that
case statement work, others don't, and I do not recognize any correlation in
the values. Commenting out any one of the case statements also works.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 11 20:43:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 11 Jul 2009 20:43:04 -0500
Subject: [LLVMbugs] [Bug 4541] New: crash with -g
Message-ID: <bug-4541-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4541

           Summary: crash with -g
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


The attached .mi crashes when compiled with debug info. The problem looks like
a double free, and it is very susceptible to changes in the input. valgrind may
come in handy.
--
ddunbar at giles:delta$ gcc -w -c t.mi
ddunbar at giles:delta$ clang -w -c t.mi
ddunbar at giles:delta$ clang -w -g -c t.mi
0   clang-cc          0x0000000100bbda15 PrintStackTrace(void*) + 38
1   clang-cc          0x0000000100bbdf25 SignalHandler(int) + 288
2   libSystem.B.dylib 0x00007fff84b3c2fa _sigtramp + 26
3   libSystem.B.dylib 0x0000000102711ab8 _sigtramp + 2109560792
4   clang-cc          0x00000001007c9bb0 llvm::PATypeHolder::get() const + 24
5   clang-cc          0x00000001007c9bf5 llvm::PATypeHolder::operator
llvm::Type*() const + 21
6   clang-cc          0x00000001007c9c11 llvm::Value::getType() const + 25
7   clang-cc          0x0000000100adebca
llvm::ConstantExpr::getBitCast(llvm::Constant*, llvm::Type const*) + 26
8   clang-cc          0x0000000100b32af1
llvm::LLVMContext::getConstantExprBitCast(llvm::Constant*, llvm::Type const*) +
33
9   clang-cc          0x0000000100a6225b
llvm::DIFactory::getCastToEmpty(llvm::DIDescriptor) + 85
10  clang-cc          0x0000000100a635f5
llvm::DIFactory::CreateDerivedType(unsigned int, llvm::DIDescriptor,
std::string const&, llvm::DICompileUnit, unsigned int, unsigned long long,
unsigned long long, unsigned long long, unsigned int, llvm::DIType) + 405
11  clang-cc          0x00000001001002e1
clang::CodeGen::CGDebugInfo::CreateType(clang::ObjCInterfaceType const*,
llvm::DICompileUnit) + 897
12  clang-cc          0x00000001000fee4e
clang::CodeGen::CGDebugInfo::getOrCreateType(clang::QualType,
llvm::DICompileUnit) + 566
13  clang-cc          0x00000001000ff7fb
clang::CodeGen::CGDebugInfo::EmitDeclare(clang::VarDecl const*, unsigned int,
llvm::Value*, llvm::IRBuilder<true, llvm::ConstantFolder>&) + 229
14  clang-cc          0x00000001000ff989
clang::CodeGen::CGDebugInfo::EmitDeclareOfAutoVariable(clang::VarDecl const*,
llvm::Value*, llvm::IRBuilder<true, llvm::ConstantFolder>&) + 53
15  clang-cc          0x00000001001067e6
clang::CodeGen::CodeGenFunction::EmitLocalBlockVarDecl(clang::VarDecl const&) +
1828
16  clang-cc          0x0000000100107771
clang::CodeGen::CodeGenFunction::EmitBlockVarDecl(clang::VarDecl const&) + 157
17  clang-cc          0x00000001001078da
clang::CodeGen::CodeGenFunction::EmitDecl(clang::Decl const&) + 310
18  clang-cc          0x0000000100154d66
clang::CodeGen::CodeGenFunction::EmitDeclStmt(clang::DeclStmt const&) + 60
19  clang-cc          0x00000001001557a6
clang::CodeGen::CodeGenFunction::EmitStmt(clang::Stmt const*) + 512
20  clang-cc          0x0000000100156871
clang::CodeGen::CodeGenFunction::EmitCompoundStmt(clang::CompoundStmt const&,
bool, llvm::Value*, bool) + 301
21  clang-cc          0x0000000100156a76
clang::CodeGen::CodeGenFunction::EmitSimpleStmt(clang::Stmt const*) + 190
22  clang-cc          0x00000001001555f4
clang::CodeGen::CodeGenFunction::EmitStmt(clang::Stmt const*) + 78
23  clang-cc          0x000000010012a2e1
clang::CodeGen::CodeGenFunction::GenerateObjCMethod(clang::ObjCMethodDecl
const*) + 161
24  clang-cc          0x000000010016489c
clang::CodeGen::CodeGenModule::EmitTopLevelDecl(clang::Decl*) + 892
25  clang-cc          0x000000010017693a (anonymous
namespace)::CodeGeneratorImpl::HandleTopLevelDecl(clang::DeclGroupRef) + 76
26  clang-cc          0x000000010003c14d (anonymous
namespace)::BackendConsumer::HandleTopLevelDecl(clang::DeclGroupRef) + 139
27  clang-cc          0x00000001002378b4 clang::ParseAST(clang::Preprocessor&,
clang::ASTConsumer*, clang::ASTContext&, bool, bool) + 478
28  clang-cc          0x000000010002871e ProcessInputFile(clang::Preprocessor&,
clang::PreprocessorFactory&, std::string const&, ProgActions,
llvm::StringMap<bool, llvm::MallocAllocator> const&, llvm::LLVMContext&) + 4175
29  clang-cc          0x000000010002a0aa main + 2877
30  clang-cc          0x0000000100024680 start + 52
31  clang-cc          0x0000000000000017 start + 4294818251
Stack dump:
0.      Program arguments: /Users/ddunbar/llvm/Debug/bin/clang-cc -triple
x86_64-apple-darwin10 -S -disable-free -main-file-name t.mi --relocation-model
pic -pic-level=1 --disable-fp-elim --unwind-tables=1 --mcpu=core2
--fmath-errno=0 -mmacosx-version-min=10.6.0 -g -w -fdiagnostics-show-option -o
/var/folders/DQ/DQ8GT3++HESEzT1obWBynE+++TI/-Tmp-/cc-e8Z42k.s -x
objective-c-cpp-output t.mi 
1.      t.mi:14:1: current parser token '@'
2.      t.mi:11:1: LLVM IR generation of declaration
'CYObjectsController::attributedStringWithString:image:'
3.      t.mi:11:87: LLVM IR generation of compound statement ('{}')
clang: error: compiler command failed due to signal 11 (use -v to see
invocation)
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 03:56:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 03:56:05 -0500
Subject: [LLVMbugs] [Bug 4542] New: target-dependent optimization with
 unspecified target
Message-ID: <bug-4542-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4542

           Summary: target-dependent optimization with unspecified target
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ubub at gmx.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3183)
 --> (http://llvm.org/bugs/attachment.cgi?id=3183)
example

hi,
while migrating my program that utilizes the llvm toolchain from version 2.3 to
2.5, opt introduces a optimization of GEPs that seems to be target-dependent,
but neither the target is specified in the IR nor on the command-line nor does
it match the pointer-size of the current platform.

I'm using the precompiled "LLVM Binaries for Mingw32/x86" on 32-bit Windows.

As an example I've attached the source IR and the optimized version with and
without a target datalayout.


best regards
tobias


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 14:39:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 14:39:18 -0500
Subject: [LLVMbugs] [Bug 4543] New: remove llvm.part.select and
	llvm.part.set intrinsics
Message-ID: <bug-4543-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4543

           Summary: remove llvm.part.select and llvm.part.set intrinsics
           Product: libraries
           Version: 1.0
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


These have never been implemented in-tree and are better handled with pattern
matching anyway, they should just be removed.  No autoupgrade support is needed
because no front-end has ever generated them.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 15:05:30 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 15:05:30 -0500
Subject: [LLVMbugs] [Bug 4113] on-the-fly FunctionPass that requires
 TargetData: Assertion `0 && "ERROR: Bad TargetData ctor used. " " Tool did
 not specify a TargetData to use?"' failed.
In-Reply-To: <bug-4113-206@http.llvm.org/bugs/>
Message-ID: <200907122005.n6CK5UAo030089@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4113


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-07-12 15:05:29 ---


*** This bug has been marked as a duplicate of bug 4112 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 15:18:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 15:18:58 -0500
Subject: [LLVMbugs] [Bug 4544] New: regalloc problem with insert_subreg
Message-ID: <bug-4544-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4544

           Summary: regalloc problem with insert_subreg
           Product: libraries
           Version: 1.0
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


It looks like regalloc is incorrectly updating machine instrs after coalescing.
 With the attached patch on the attached .bc file, I get:

$ llc bugpoint-reduced-simplified.bc -o - -print-machineinstrs
...
        %reg1039<def> = MOV32r0 %EFLAGS<imp-def>
        %reg1040<def> = IMPLICIT_DEF
        %reg1037<def> = INSERT_SUBREG %reg1040, %reg1039, 4
..
        %RSI<def> = MOV64rr %reg1037
        %RDX<def> = MOV64rr %reg1037
        %RCX<def> = MOV64rr %reg1037

The after regalloc:

        %ESI<def> = MOV32r0 %EFLAGS<imp-def,dead>
        %RDX<def> = MOV32ri 0
        %RCX<def> = MOV32ri 0
        %R8<def> = MOV64rr %RAX<kill>

The set of ESI seems fine.  The set of RDX/RCX are wrong (they should be
EDX/ECX).

This smells like a simple remat problem, or an "is cheap as a move" problem of
some sort.  It would also be nice to use MOV32ri0 for EDX/ECX instead of
MOV32ri, but that's a micro optimization possible because the flags are dead.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 16:09:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 16:09:14 -0500
Subject: [LLVMbugs] [Bug 4543] remove llvm.part.select and llvm.part.set
	intrinsics
In-Reply-To: <bug-4543-206@http.llvm.org/bugs/>
Message-ID: <200907122109.n6CL9EVq032495@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4543


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-12 16:09:14 ---
Done here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090706/080814.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 12 21:59:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 12 Jul 2009 21:59:09 -0500
Subject: [LLVMbugs] [Bug 4545] New: ConstantRange::multiply isn't always
	conservative
Message-ID: <bug-4545-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4545

           Summary: ConstantRange::multiply isn't always conservative
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: nicholas at mxc.ca, llvmbugs at cs.uiuc.edu,
                    cristic at stanford.edu


ConstantRange::multiply sometimes computes a set which does not include some
possible values.

Here are some klee generated test cases:
ERROR: Found invalid values!
  A = 1 is in LHS = [1, 0), and
  B = 0 is in RHS = [0, 2), but
  A mul B = 0 is not in (LHS mul RHS) = [0, 0)
ERROR: Found invalid values!
  A = 1 is in LHS = [0, 2), and
  B = 1 is in RHS = [1, 0), but
  A mul B = 1 is not in (LHS mul RHS) = [0, 0)
ERROR: Found invalid values!
  A = 2 is in LHS = [1, 6), and
  B = 7 is in RHS = [6, 2), but
  A mul B = 14 is not in (LHS mul RHS) = [6, 6)
ERROR: Found invalid values!
  A = 1 is in LHS = [1, 6), and
  B = 7 is in RHS = [6, 2), but
  A mul B = 7 is not in (LHS mul RHS) = [6, 6)
ERROR: Found invalid values!
  A = 10 is in LHS = [6, 14), and
  B = 11 is in RHS = [11, 6), but
  A mul B = 14 is not in (LHS mul RHS) = [2, 2)
ERROR: Found invalid values!
  A = 2 is in LHS = [2, 8), and
  B = 4 is in RHS = [4, 2), but
  A mul B = 8 is not in (LHS mul RHS) = [8, 8)
ERROR: Found invalid values!
  A = 3 is in LHS = [1, 0), and
  B = 3 is in RHS = [2, 0), but
  A mul B = 9 is not in (LHS mul RHS) = [2, 2)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 00:31:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 00:31:13 -0500
Subject: [LLVMbugs] [Bug 4546] New: Implement warning for irrelevant switch
	statements
Message-ID: <bug-4546-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4546

           Summary: Implement warning for irrelevant switch statements
           Product: clang
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


MSVC has a nice warning for code like
--
switch (foo) {
}
--
or 
--
switch (foo) {
default:
...
}
--
which would be nice to have in clang (if only to improve portability for
clients who would like their code to be warning free under MSVC).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 11:05:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 11:05:36 -0500
Subject: [LLVMbugs] [Bug 4548] New: missed optimization: zext 8/16->32+udiv
 32+trunc 32->8/16 - > udiv 8/16
Message-ID: <bug-4548-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4548

           Summary: missed optimization: zext 8/16->32+udiv 32+trunc 32-
                    >8/16 -> udiv 8/16
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


For the C code below:

uint8_t udiv_i8(uint8_t a, uint8_t b)
{
  return a / b;
}

Clang generates this IR (at -O1/O2):

        %conv = zext i8 %a to i32       
        %conv2 = zext i8 %b to i32      
        %div = udiv i32 %conv, %conv2   
        %conv3 = trunc i32 %div to i8   
        ret i8 %conv3

This should be optimized to this IR, since that  is what the original C code
meant:

%div = udiv i8 %a, %b
ret i8 %div


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 12:03:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 12:03:55 -0500
Subject: [LLVMbugs] [Bug 4541] crash with -g
In-Reply-To: <bug-4541-206@http.llvm.org/bugs/>
Message-ID: <200907131703.n6DH3tKq023203@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4541


devang.patel <devang.patel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|dpatel at apple.com            |devang.patel at gmail.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from devang.patel <devang.patel at gmail.com>  2009-07-13 12:03:55 ---
fixed in rev. 75473.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 12:33:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 12:33:46 -0500
Subject: [LLVMbugs] [Bug 4256] MemoryDependenceAnalysis.cpp:93: Assertion
 `InstIt != ReverseMap .end() && "Reverse map out of sync?"' failed
In-Reply-To: <bug-4256-206@http.llvm.org/bugs/>
Message-ID: <200907131733.n6DHXkgG025395@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4256


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-13 12:33:44 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/080851.html

Thanks to Jakub Staszak for tracking down the bug and putting together the
patch.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 12:42:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 12:42:31 -0500
Subject: [LLVMbugs] [Bug 4549] New: Incompatible input / output inline asm
 constraints in glibc code
Message-ID: <bug-4549-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4549

           Summary: Incompatible input / output inline asm constraints in
                    glibc code
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: llvmbugs at cs.uiuc.edu


Consider the following C code:

#include <stdlib.h>

void foo(void) {
  fd_set fdbar;
  FD_ZERO(&fdbar);
}

llvm-gcc emits the following error:

1.c: In function 'foo':
1.c:5: error: unsupported inline asm: input constraint with a matching output
constraint of incompatible type!

This break Qt and (I assume) many other apps. Preprocessed source is attached.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 13:01:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 13:01:31 -0500
Subject: [LLVMbugs] [Bug 4549] Incompatible input / output inline asm
 constraints in glibc code
In-Reply-To: <bug-4549-206@http.llvm.org/bugs/>
Message-ID: <200907131801.n6DI1V7a027385@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4549


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dalej at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #2 from Dale Johannesen <dalej at apple.com>  2009-07-13 13:01:29 ---
This fails for me only in 64-bit mode; I assume that's what you're using.  The
asm is matching a 32-bit int output with a 64-bit pointer input.  My
understanding is Chris doesn't want to support this.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 13:05:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 13:05:42 -0500
Subject: [LLVMbugs] [Bug 4549] Incompatible input / output inline asm
 constraints in glibc code
In-Reply-To: <bug-4549-206@http.llvm.org/bugs/>
Message-ID: <200907131805.n6DI5gw0027659@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4549


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-07-13 13:05:41 ---
this is static x86-64 mode, which darwin doesn't support.  This should be
legal.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 17:12:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 17:12:25 -0500
Subject: [LLVMbugs] [Bug 4522] cmake builds uses configure builds tblgen
 output instead of generating its own
In-Reply-To: <bug-4522-206@http.llvm.org/bugs/>
Message-ID: <200907132212.n6DMCPvm003724@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4522


??scar Fuentes <ofv at wanadoo.es> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from ??scar Fuentes <ofv at wanadoo.es>  2009-07-13 17:12:24 ---
r75510 is the non-optimal fix for the missing dependencies of .td files, as
described on my previous entry.

On r75522 cmake checks the existence of tablegenned files mixed with the
sources. This only works when cmake is executed. This check should make
unnecessary to prepend the list of include paths with the current obj
directory. (The converse is true: prepending the list of include paths with the
current obj directory makes the check unnecessary, and maybe it is the correct
thing to do).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 17:40:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 17:40:36 -0500
Subject: [LLVMbugs] [Bug 4534] LLVM taking too long to build a file
In-Reply-To: <bug-4534-206@http.llvm.org/bugs/>
Message-ID: <200907132240.n6DMeaXb004824@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4534


revcompgeek at gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #8 from revcompgeek at gmail.com  2009-07-13 17:40:35 ---
This no longer is problem as far as I can tell with the latest llvm.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 17:50:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 17:50:49 -0500
Subject: [LLVMbugs] [Bug 4548] missed optimization: zext 8/16->32+udiv
 32+trunc 32->8/16 -> udiv 8/16
In-Reply-To: <bug-4548-206@http.llvm.org/bugs/>
Message-ID: <200907132250.n6DMonhT005221@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4548


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-13 17:50:48 ---
Fixed in r75539.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 18:08:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 18:08:14 -0500
Subject: [LLVMbugs] [Bug 4537] infinite loop in scev
In-Reply-To: <bug-4537-206@http.llvm.org/bugs/>
Message-ID: <200907132308.n6DN8E3j005821@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4537


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Dan Gohman <gohman at apple.com>  2009-07-13 18:08:14 ---
This is now fixed, here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/080895.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 18:43:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 18:43:28 -0500
Subject: [LLVMbugs] [Bug 4549] Incompatible input / output inline asm
 constraints in glibc code
In-Reply-To: <bug-4549-206@http.llvm.org/bugs/>
Message-ID: <200907132343.n6DNhSv9006995@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4549


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #12 from Anton Korobeynikov <asl at math.spbu.ru>  2009-07-13 18:43:01 ---
Fixed in

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/080945.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/080946.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 19:52:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 19:52:02 -0500
Subject: [LLVMbugs] [Bug 4544] regalloc problem with insert_subreg
In-Reply-To: <bug-4544-206@http.llvm.org/bugs/>
Message-ID: <200907140052.n6E0q2V9009395@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4544


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #9 from Evan Cheng <evan.cheng at apple.com>  2009-07-13 19:51:55 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/080965.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 21:39:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 21:39:16 -0500
Subject: [LLVMbugs] [Bug 4551] New: segfault
Message-ID: <bug-4551-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4551

           Summary: segfault
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp168$ llvm-gcc -O2 -w small.c
small.c:52: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

regehr at john-home:~/volatile/tmp168$ llvm-gcc -v

Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../configure
--prefix=/home/regehr/z/tmp/llvm-gcc-r75580-install
--program-prefix=llvm-r75580- --enable-languages=c,c++
--enable-llvm=/home/regehr/z/tmp/llvm-r75580 --enable-checking=release
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5646) (LLVM build)

regehr at john-home:~/volatile/tmp168$ cat small.c

char
safe_mul_func_char_u_u (char _ui1, char _ui2)
{
  return _ui1;
}

char
safe_add_func_int_u_u (int _ui1, char _ui2)
{
  return _ui1 + 1;
}

char
safe_sub_func_int_u_u (int _ui1, char _ui2)
{
  return _ui1 - 1;
}

char
safe_mod_func_int_u_u (int _ui1, char _ui2)
{
  return 1 ? _ui1 : 1;
}
struct S1
{
};
int g_11;
char g_64;
struct S1 func_53 (int p_54, int p_55);
char
func_16 (int p_17)
{
  return p_17;
}

uint87 (int p_28, char p_29, char p_31, char p_32)
{
  char l_34;
  for (1; l_34; l_34 = safe_sub_func_int_u_u (l_34, 1))
    {
      func_53 (p_32, !g_64);
    }
}
struct S1
func_53 (int p_54, int p_55)
{
  for (p_55 = 1; p_55; p_55 = safe_add_func_int_u_u (p_55, 1))
    g_11 =
      safe_mod_func_int_u_u (func_16
                                  (safe_mul_func_char_u_u (p_55, 1)),
                                  func_16);
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 21:52:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 21:52:07 -0500
Subject: [LLVMbugs] [Bug 4552] New: Error: bad register name `%dil'
Message-ID: <bug-4552-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4552

           Summary: Error: bad register name `%dil'
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp168$  llvm-gcc -O3 small.c
/tmp/ccApvYXi.s: Assembler messages:
/tmp/ccApvYXi.s:184: Error: bad register name `%dil'

regehr at john-home:~/volatile/tmp168$ llvm-gcc -v

Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../configure
--prefix=/home/regehr/z/tmp/llvm-gcc-r75580-install
--program-prefix=llvm-r75580- --enable-languages=c,c++
--enable-llvm=/home/regehr/z/tmp/llvm-r75580 --enable-checking=release
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5646) (LLVM build)

regehr at john-home:~/volatile/tmp168$ cat small.c

char
safe_mod_func_char_s_s (char _si1, char _si2)
{
  return 1 ? _si1 : +1;
}

char
safe_mod_func_int_s_s (int _si1, char _si2)
{
  return -32767 - 1 ? _si1 : +1;
}

char
safe_add_func_char_u_u (char _ui1, char _ui2)
{
}
char
safe_mul_func_char_u_u (char _ui1, char _ui2)
{
  return _ui1 * _ui2;
}

int g_8;
int g_72;
uint84 (int p_15, char p_17, char p_19)
{
  for (1; 1; safe_add_func_char_u_u)
    {
      func_40 (1);
      for (1; g_8 <= 0; g_8 = 1)
        {
        }
    }
}
int
func_21 (int16p_22)
{
  int l_23 = 0xC477L;
  return l_23;
}

int
func_40 (int p_41, int p_42, int p_43, int p_45)
{
  func_55 (~1);
  func_55 (1);
}

char
func_51 (int64p_52)
{
  return g_8;
}

int
func_55 (int16p_56)
{
  if (safe_mod_func_int_s_s
      (safe_mul_func_char_u_u (func_51 (1), func_70 (!1)), 0))
    if (func_21 (1) & safe_mod_func_char_s_s (g_72, 0))
      return 1;
  g_72 = 1;
}

int
func_70 (int p_71, int p_73)
{
  char l_74;
  l_74 = g_72;
  return l_74;
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 13 21:58:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 13 Jul 2009 21:58:25 -0500
Subject: [LLVMbugs] [Bug 4553] New: Assertion `NodePtr && "--'d off the
 beginning of an ilist!" ' failed.
Message-ID: <bug-4553-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4553

           Summary: Assertion `NodePtr && "--'d off the beginning of an
                    ilist!"' failed.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen using Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp168$ llvm-gcc -O small.c
small.c:25: warning: large integer implicitly truncated to unsigned type
small.c:28: warning: excess elements in struct initializer
small.c:28: warning: (near initialization for ???g_180???)
small.c: In function ???func_167???:
small.c:41: warning: excess elements in struct initializer
small.c:41: warning: (near initialization for ???l_208.f2.f1???)
cc1: /home/regehr/z/tmp/llvm-r75580/include/llvm/ADT/ilist.h:213:
llvm::ilist_iterator<NodeTy>& llvm::ilist_iterator<NodeTy>::operator--() [with
NodeTy = llvm::Instruction]: Assertion `NodePtr && "--'d off the beginning of
an ilist!"' failed.
small.c: At top level:
small.c:47: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

regehr at john-home:~/volatile/tmp168$ llvm-gcc -v

Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../configure
--prefix=/home/regehr/z/tmp/llvm-gcc-r75580-install
--program-prefix=llvm-r75580- --enable-languages=c,c++
--enable-llvm=/home/regehr/z/tmp/llvm-r75580 --enable-checking=release
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5646) (LLVM build)

regehr at john-home:~/volatile/tmp168$ cat small.c

unsigned short safe_sub_func_uint_u_u (unsigned short _ui1, unsigned short
_ui2)
{
  return -_ui2;
}

struct S0
{
};

struct S1
{
  unsigned short f0;
  struct S0 f1;
};

struct S2
{
  char f1;
  struct S1 f2;
};

int g_143;

struct S2 g_180 = {
  -3L, 0xB066DA58L, {
  }
  , 0x6ED36738L
};

char func_24 (struct S2 p_25, char p_26, char p_27)
{
  return p_25.f2.f0;
}

struct S0 func_167 (struct S1 p_168, char p_169)
{
  for (g_143 = 0; g_143 >= 0; g_143 = safe_sub_func_uint_u_u (g_143, 1))
    {
      struct S2 l_208 = {
        0, 0L, 0xC23F51DEL
      };
      for (g_143 = 0; 0; 1)
        {
        }
      p_168.f0 = func_24 (l_208, g_180.f1, 1);
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 05:43:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 05:43:44 -0500
Subject: [LLVMbugs] [Bug 4555] New: Doesn't support valid obj-c declaration:
 -(Class<MyClass>) currentCounterPolicy
Message-ID: <bug-4555-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4555

           Summary: Doesn't support valid obj-c declaration: -
                    (Class<MyClass>) currentCounterPolicy
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: major
          Priority: P2
         Component: parser
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: macta at pobox.com
                CC: llvmbugs at cs.uiuc.edu


XCode 3.1.x correctly compiles the following (only snippets shown - but I think
the idea should be clear. Basically it provides a hierarchy of Policy counter
objects, classes are used as singleton objects - each with a different static
implementation):

Clang gives the following error:
Model.h:38:22: error: protocol qualified 'Class' is unsupported
-(Class<CounterPolicy>) currentCounterPolicy;
  ~~~~~~~~~~~~~~~~~~~^
1 diagnostic generated.


Code snippets:

Model.h

...
#import "ModelPolicies.h"
...

@interface Model : NSObject {
// inst var declarations etc.
...

// The line Clang reports an error on
-(Class<CounterPolicy>) currentCounterPolicy; 


Model.m
....

// Create a static list of counter policy objects
+(NSArray *) availableCounters {
        return [[[NSArray alloc] initWithObjects: [CurrentListCounter class],
[FavouritesCounter class], [AllListsCounter class], [NullCounter class], nil]
autorelease];

}

-(Class<CounterPolicy>) currentCounterPolicy {
        return [[Model availableCounters] objectAtIndex:
self.badgeCounterIndex];
}



ModelPolicies.h

@protocol CounterPolicy

+(NSInteger) countFor: (MyModel *)model;
+(NSString *) description;
@end

@interface CurrentListCounter : NSObject <CounterPolicy> 

+(NSInteger) badgeCountFor: (MyModel *)model;
+(NSString *) description;

@end

@interface NullCounter : NSObject <CounterPolicy> 

+(NSInteger) badgeCountFor: (MyModel *)model;
+(NSString *) description;

@end


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 06:53:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 06:53:08 -0500
Subject: [LLVMbugs] [Bug 4556] New: Assertion initialising global with
 address of " extern void foo"
Message-ID: <bug-4556-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4556

           Summary: Assertion initialising global with address of "extern
                    void foo"
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: richard at xmos.com
                CC: llvmbugs at cs.uiuc.edu


With the following code:

extern void foo;
void *bar = &foo;

llvm-gcc built from trunk asserts:

$ ./cc1 test.c -emit-llvm 
cc1: /local/richard/llvm/lib/VMCore/Type.cpp:1327: static llvm::PointerType*
llvm::PointerType::get(const llvm::Type*, unsigned int): Assertion `ValueType
!= Type::VoidTy && "Pointer to void is not valid, use i8* instead!"' failed.
test.c:2: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

This code compiles without error using the 2.5 release.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 13:21:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 13:21:50 -0500
Subject: [LLVMbugs] [Bug 4553] Assertion `NodePtr && "--'d off the beginning
 of an ilist!"' failed.
In-Reply-To: <bug-4553-206@http.llvm.org/bugs/>
Message-ID: <200907141821.n6EILok3025963@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4553


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Dan Gohman <gohman at apple.com>  2009-07-14 13:21:43 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081024.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 13:22:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 13:22:05 -0500
Subject: [LLVMbugs] [Bug 4557] New: differences with -tradtional compared to
	gcc
Message-ID: <bug-4557-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4557

           Summary: differences with -tradtional compared to gcc
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: NetBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: krister.walfridsson at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3197)
 --> (http://llvm.org/bugs/attachment.cgi?id=3197)
The example mentioned in the bug report

"clang -traditional" gives different output compared to "gcc -traditional" on
the following example. (This prevents the NetBSD kernel from building using
clang, as this construct is used in several .S files...)

Example:
#define INTRSTUB(name, num) \
intr_/**/name/**/num: \
        nop
INTRSTUB(legacy,0)


"clang -traditional -E foo.S" outputs this as "intr_ legacy 0: nop"
"gcc -traditional -E foo.S" outputs this as "intr_legacy0: nop"

The clang version used: svn revision 75635.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 13:22:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 13:22:25 -0500
Subject: [LLVMbugs] [Bug 4551] segfault
In-Reply-To: <bug-4551-206@http.llvm.org/bugs/>
Message-ID: <200907141822.n6EIMPCX026004@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4551


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Dan Gohman <gohman at apple.com>  2009-07-14 13:22:22 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081024.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 13:43:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 13:43:55 -0500
Subject: [LLVMbugs] [Bug 4556] Assertion initialising global with address of
 "extern void foo"
In-Reply-To: <bug-4556-206@http.llvm.org/bugs/>
Message-ID: <200907141843.n6EIhtaq026878@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4556


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-14 13:43:49 ---
Fixed, thanks:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081055.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 14 21:47:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 14 Jul 2009 21:47:07 -0500
Subject: [LLVMbugs] [Bug 3515] llc build with Visual studio 2008 assert for
 not link X86AsmPrinter
In-Reply-To: <bug-3515-206@http.llvm.org/bugs/>
Message-ID: <200907150247.n6F2l7nn012315@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3515


StanLi <stanli at S3Graphics.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 07:58:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 07:58:28 -0500
Subject: [LLVMbugs] [Bug 4535] mark constant strings private instead of
	internal
In-Reply-To: <bug-4535-206@http.llvm.org/bugs/>
Message-ID: <200907151258.n6FCwSo9014497@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4535


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2009-07-15 07:57:36 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081177.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 09:36:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 09:36:38 -0500
Subject: [LLVMbugs] [Bug 4559] New: llvm-gcc build broken: ident string in
	middle of .s
Message-ID: <bug-4559-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4559

           Summary: llvm-gcc build broken: ident string in middle of .s
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


On x86-64 linux I am seeing a llvm-gcc build failure when xgcc compiles
_lshrdi3,
part of libgcc:

/tmp/ccvJKDEb.s: Assembler messages:
/tmp/ccvJKDEb.s:355: Error: junk at end of line, first unrecognized character
is `.'

What's happening is that the ident string has turned up in the middle of
the file (at line 355):

        .byte   0x2
        .quad   .Llabel7
        .byte   0x16    .ident  "GCC: (GNU) 4.2.1 (Based on Apple Inc. build
5646) (LLVM build)"


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 11:23:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 11:23:12 -0500
Subject: [LLVMbugs] [Bug 4560] New: "Recursive compilation detected" and
	Linux signals
Message-ID: <bug-4560-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4560

           Summary: "Recursive compilation detected" and Linux signals
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: mattp223 at renzelmann.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3199)
 --> (http://llvm.org/bugs/attachment.cgi?id=3199)
C program that should crash lli utility with "Recursive compilation detected"
error

I reported this to the llvmdev mailing list yesterday because bugzilla was not
responding for some reason, but everything appears fine today, so I thought I'd
post here so it'd be more accessible.

Platform is RHEL5, GCC 4.2.4, x86-32, and LLVM/LLVM-GCC from subversion
(two days ago).  I'm compiling C code into bitcode, and then executing
the bitcode using the JIT compiler (lli).

I've managed to reproduce a problem when multiple signals go off around the
same time.  A small sample program is attached.  The result is the "recursive
compilation detected" JIT compiler error.  The same program compiled via GCC
works fine.  Note that the sample program uses the preprocessor to create
large amounts of superfluous code to tie-up the JIT compiler.

Compile the attachment with:
llvm-gcc -c -emit-llvm ./main.c -o main.bc

And run it with:
lli main.bc

Here's the output from the debug version of the lli utility.  Note that I
slightly tweaked my version of LLVM to print out the function that it's
attempting to JIT, and the function that it was previously JIT'ing when it
detected the recursion.  This print statement makes it clear that the JIT
compiler was working on sigsegv_handler before it attempted to start on
timer_handler.
=====================================
Recursive compilation:
Function: timer_handler, existing function sigsegv_handler
lli: JIT.cpp:579: void llvm::JIT::runJITOnFunctionUnlocked(llvm::Function*,
const llvm::MutexGuard&): Assertion `!isAlreadyCodeGenerating && "Error:
Recursive compilation detected!"' failed.
0   lli       0x08a5308e
1   lli       0x08a53601
2             0x008e0420 __kernel_sigreturn + 0
3   libc.so.6 0x00146ac1 abort + 257
4   libc.so.6 0x0013e39e __assert_fail + 238
5   lli       0x08738aa4
llvm::JIT::runJITOnFunctionUnlocked(llvm::Function*, llvm::MutexGuard
const&) + 408
6   lli       0x08738f6d llvm::JIT::getPointerToFunction(llvm::Function*) +
755
7   lli       0x0873f704
8   lli       0x08597416 X86CompilationCallback2 + 166
9   lli       0x08596e08 X86CompilationCallback + 24
10  lli       0x00eebff5 X86CompilationCallback + 4170535429
11  lli       0x0863e13e llvm::CallSDNode::CallSDNode(unsigned int,
llvm::DebugLoc, bool, bool, bool, llvm::SDVTList, llvm::SDValue const*,
unsigned int, unsigned int) + 90
12  lli       0x08622d47 llvm::SelectionDAG::getCall(unsigned int,
llvm::DebugLoc, bool, bool, bool, llvm::SDVTList, llvm::SDValue const*,
unsigned int, unsigned int) + 419
13  lli       0x08648853 llvm::TargetLowering::LowerCallTo(llvm::SDValue,
llvm::Type const*, bool, bool, bool, bool, unsigned int, unsigned int, bool,
llvm::SDValue, std::vector<llvm::TargetLowering::ArgListEntry,
std::allocator<llvm::TargetLowering::ArgListEntry> >&, llvm::SelectionDAG&,
llvm::DebugLoc) + 2125
14  lli       0x0864f8c6
llvm::SelectionDAGLowering::LowerCallTo(llvm::CallSite, llvm::SDValue, bool,
llvm::MachineBasicBlock*) + 1418
15  lli       0x0865c6b1
llvm::SelectionDAGLowering::visitCall(llvm::CallInst&) + 2527
16  lli       0x08666d1e llvm::SelectionDAGLowering::visit(unsigned int,
llvm::User&) + 1108
17  lli       0x08666e16
llvm::SelectionDAGLowering::visit(llvm::Instruction&) + 52
18  lli       0x086824ff
llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*,
llvm::ilist_iterator<llvm::Instruction>,
llvm::ilist_iterator<llvm::Instruction>) + 99
19  lli       0x086830fd
llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&,
llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*,
llvm::TargetInstrInfo const&) + 2433
20  lli       0x08683f20
llvm::SelectionDAGISel::runOnFunction(llvm::Function&) + 908
21  lli       0x089dad71 llvm::FPPassManager::runOnFunction(llvm::Function&)
+ 289
22  lli       0x089db302 llvm::FunctionPassManagerImpl::run(llvm::Function&)
+ 124
23  lli       0x089db4d1 llvm::FunctionPassManager::run(llvm::Function&) +
135
24  lli       0x08738b09
llvm::JIT::runJITOnFunctionUnlocked(llvm::Function*, llvm::MutexGuard
const&) + 509
25  lli       0x08738f6d llvm::JIT::getPointerToFunction(llvm::Function*) +
755
26  lli       0x0873f704
27  lli       0x08597416 X86CompilationCallback2 + 166
28  lli       0x08596e08 X86CompilationCallback + 24
29  lli       0x00eebfed X86CompilationCallback + 4170535421
30  lli       0x08759bdd
llvm::ExecutionEngine::runFunctionAsMain(llvm::Function*,
std::vector<std::string, std::allocator<std::string> > const&, char const*
const*) + 1255
31  lli       0x08390e66 main + 1946
32  libc.so.6 0x001316d0 __libc_start_main + 224
33  lli       0x0838fd61
Stack dump:
0.      Program arguments: /scratch/sym/llvm-current/Debug/bin/lli main.bc 
1.      Running pass 'X86 DAG->DAG Instruction Selection' on function
'@sigsegv_handler'
Aborted
=====================================

Please let me know if I'm missing something :)  Is this behavior by design
or is it a bug/limitation?

Thanks and regards,
Matt


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 12:45:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 12:45:19 -0500
Subject: [LLVMbugs] [Bug 2700] LLVM silently ignores I/O errors
In-Reply-To: <bug-2700-206@http.llvm.org/bugs/>
Message-ID: <200907151745.n6FHjJi8027593@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2700


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Dan Gohman <gohman at apple.com>  2009-07-15 12:45:18 ---
This is now fixed in r75758 and r75758.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 17:43:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 17:43:26 -0500
Subject: [LLVMbugs] [Bug 3265] You can' t know the number of arguments from
 the non_null attribute.
In-Reply-To: <bug-3265-206@http.llvm.org/bugs/>
Message-ID: <200907152243.n6FMhQnI007361@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3265


Ted Kremenek <kremenek at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |INVALID




--- Comment #2 from Ted Kremenek <kremenek at apple.com>  2009-07-15 17:42:30 ---
Sorry for the delay in revisiting this bug.  I realized after I looked at this
a little more closely that changing NonNullAttr is not necessary.  From a
FunctionDecl one can iterate through its declarations to find one that has the
arguments in place.  Modifying NonNullAttr would actually be a mistake, as it
shouldn't cache information about the FunctionDecl that can easily be garnered
from the FunctionDecl itself.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 20:44:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 20:44:12 -0500
Subject: [LLVMbugs] [Bug 4559] llvm-gcc build broken: ident string in middle
	of .s
In-Reply-To: <bug-4559-206@http.llvm.org/bugs/>
Message-ID: <200907160144.n6G1iCqG013611@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4559


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gohman at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #8 from Dan Gohman <gohman at apple.com>  2009-07-15 20:44:11 ---
This is fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081295.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 22:49:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 22:49:53 -0500
Subject: [LLVMbugs] [Bug 3982] reject attribute weak on internal variables
In-Reply-To: <bug-3982-206@http.llvm.org/bugs/>
Message-ID: <200907160349.n6G3nr2p018093@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3982


Ryan Flynn <parseerror at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Ryan Flynn <parseerror at gmail.com>  2009-07-15 22:49:41 ---
Fixed in r75879. Testcase added to test/Sema/attr-weak.c


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 15 23:55:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 15 Jul 2009 23:55:27 -0500
Subject: [LLVMbugs] [Bug 4562] New: Miscompliation of GVN
Message-ID: <bug-4562-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4562

           Summary: Miscompliation of GVN
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: haohui.mai at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Here is the reduced test case produced by Bugpoint.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 16 11:45:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 16 Jul 2009 11:45:26 -0500
Subject: [LLVMbugs] [Bug 4564] New: Add configure magic to install gold
 plugin where bfd can find it
Message-ID: <bug-4564-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4564

           Summary: Add configure magic to install gold plugin where bfd can
                    find it
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-ar
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


Configure should detect if the platform has an ar/nm/ld etc. that supports
plugins, and install that gold plugin into the appropriate location if so. This
would allow fully transparent support for LLVM .bc files on Linux with the
appropriate tool chain.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 16 13:34:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 16 Jul 2009 13:34:20 -0500
Subject: [LLVMbugs] [Bug 4565] New: ARM doesn't recognize aapcs-defined
 register synonyms
Message-ID: <bug-4565-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4565

           Summary: ARM doesn't recognize aapcs-defined register synonyms
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu


AAPCS 5.1.1 defines a1..a4 and v1..v8 as synonyms for arm registers. These
aren't recognized by LLVM causing an llvm-gcc bootstrap failure on that target.

Testcase:

void test() {
  unsigned long BEG = 0;
  unsigned long END = 1;

  register unsigned long _beg __asm ("a1") = (unsigned long) (BEG);     
  register unsigned long _end __asm ("a2") = (unsigned long) (END);     
  register unsigned long _flg __asm ("a3") = 0;                         
  register unsigned long _scno __asm ("r7") = 0xf0002;                  
  __asm __volatile ("swi 0x9f0002               @ sys_cacheflush"       
                    : "=r" (_beg)                                       
                    : "0" (_beg), "r" (_end), "r" (_flg), "r" (_scno)); 
}

Error when building:

LLVM ERROR: llvm: error: Couldn't allocate output reg for constraint '{a1}'!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 16 16:29:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 16 Jul 2009 16:29:55 -0500
Subject: [LLVMbugs] [Bug 4566] New: Invalid asmprinting for Thumb RSB
 instruction [ generic asmprinter construction problem?]
Message-ID: <bug-4566-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4566

           Summary: Invalid asmprinting for Thumb RSB instruction [generic
                    asmprinter construction problem?]
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: llvmbugs at cs.uiuc.edu


Consider the attached bitcode. Currently we're producing the following
instructions:

.Llabel18:
        lsls r5, r2
        rsbs r2, r5
        orrs r4, r2

which is definitely wrong in Thumb1 mode. In fact, tRSB instruction has the
following assembler string: "rsb", " $dst, $src, #0", but the #0 operand is
never emitted. This seems to be due to some weirdness in asmprinter emission:
AMAsmPrinter::printInstruction() function contains 2 entries for tRSB:

  case 45:
    // tMOVi8, tMVN, tRSB
    printPredicateOperand(MI, 3);
    O << " ";
    printOperand(MI, 0);
    O << ", ";
    printOperand(MI, 2);
    EmitComments(*MI);
    O << "\n";
    return true;
    break;
  }

and

  case 22:
    // tRSB
    O << ", #0";
    EmitComments(*MI);
    O << "\n";
    return true;
    break;

note that "return true' in the first case makes it impossible to reach the
second fragment.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 16 18:04:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 16 Jul 2009 18:04:20 -0500
Subject: [LLVMbugs] [Bug 4567] New: Invalid assembly produced due to unified
 syntax emission
Message-ID: <bug-4567-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4567

           Summary: Invalid assembly produced due to unified syntax emission
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3203)
 --> (http://llvm.org/bugs/attachment.cgi?id=3203)
Test bytecode

Consider the attached bytecode.

I'm having the following assembler error:

$ arm-elf-gcc gets-chk.S.s -c -march=armv6
gets-chk.S.s: Assembler messages:
gets-chk.S.s:128: Error: lo register required -- `subs sp,#12'

This seems to be fallout from recent switch to unified syntax. This breaks
llvm-gcc build in thumb mode as well.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 16 19:11:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 16 Jul 2009 19:11:15 -0500
Subject: [LLVMbugs] [Bug 4565] ARM doesn't recognize aapcs-defined register
	synonyms
In-Reply-To: <bug-4565-206@http.llvm.org/bugs/>
Message-ID: <200907170011.n6H0BFVg014011@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4565


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Dale Johannesen <dalej at apple.com>  2009-07-16 19:11:12 ---
Should be fixed here, could someone try it on ARM?
http://llvm.org/viewvc/llvm-project?rev=76128&view=rev


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 00:34:45 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 00:34:45 -0500
Subject: [LLVMbugs] [Bug 4560] "Recursive compilation detected" and Linux
	signals
In-Reply-To: <bug-4560-206@http.llvm.org/bugs/>
Message-ID: <200907170534.n6H5Yj4I024416@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4560


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |resistor at mac.com
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-17 00:34:36 ---
It would be nice to support this, but very unlikely.  Signals can interrupt
almost anything, and you can only do async-signal-safe operations from a signal
handler.  These operations do not include malloc, for example.  I can't imagine
us supporting this in the near future.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 00:36:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 00:36:36 -0500
Subject: [LLVMbugs] [Bug 4532] llc infloops in FoldingSetImpl::RemoveNode in
	llc
In-Reply-To: <bug-4532-206@http.llvm.org/bugs/>
Message-ID: <200907170536.n6H5aaRu024510@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4532


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-07-17 00:36:33 ---
Sounds like a miscompile, we'll assume that until it repro's on another system.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 09:35:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 09:35:26 -0500
Subject: [LLVMbugs] [Bug 4569] New: integer wrong code bug
Message-ID: <bug-4569-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4569

           Summary: integer wrong code bug
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen on Ubuntu Hardy.  "2" is the right answer.

regehr at john-home:~/volatile/tmp176$ llvm-gcc -O small.c -o small
regehr at john-home:~/volatile/tmp176$ ./small
checksum = 2
regehr at john-home:~/volatile/tmp176$ llvm-gcc -O2 small.c -o small
regehr at john-home:~/volatile/tmp176$ ./small
checksum = 65282
regehr at john-home:~/volatile/tmp176$ llvm-gcc -v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../configure
--prefix=/home/regehr/z/tmp/llvm-gcc-r76067-install
--program-prefix=llvm-r76067- --enable-languages=c,c++
--enable-llvm=/home/regehr/z/tmp/llvm-r76067 --enable-checking=release
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5646) (LLVM build)
regehr at john-home:~/volatile/tmp176$ cat small.c

#include <stdint.h>
#include <limits.h>
#include <stdio.h>

static uint16_t
safe_sub_func_uint16_t_u_u (uint16_t ui1, uint16_t ui2)
{
  return
    ui1-ui2;
}

uint16_t g_22;

void func_36 (void);
void func_36 (void)
{
  uint8_t l_95;
  for (l_95 = 0; l_95 != 1; l_95 = safe_sub_func_uint16_t_u_u (l_95, 1))
    g_22 = l_95;
}

int main (void)
{
  func_36 ();
  printf ("checksum = %d\n", g_22);
  return 0;

}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 17:43:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 17:43:25 -0500
Subject: [LLVMbugs] [Bug 4552] StackSlotColoring + subreg miscompile
In-Reply-To: <bug-4552-206@http.llvm.org/bugs/>
Message-ID: <200907172243.n6HMhPlB010434@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4552


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Evan Cheng <evan.cheng at apple.com>  2009-07-17 17:43:24 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081731.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 19:10:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 19:10:29 -0500
Subject: [LLVMbugs] [Bug 3774] clang: ANALYZER_STORE_MODEL=region cast
	failures
In-Reply-To: <bug-3774-206@http.llvm.org/bugs/>
Message-ID: <200907180010.n6I0ATHd013364@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3774


Ted Kremenek <kremenek at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Ted Kremenek <kremenek at apple.com>  2009-07-17 19:10:20 ---
As of r76262 one can analyze ClamAV using '-store region' with no crashes.  
Thanks for providing this excellent test case!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 20:56:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 20:56:47 -0500
Subject: [LLVMbugs] [Bug 4572] New: assert building libstdc++ for
	i686-pc-mingw32
Message-ID: <bug-4572-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4572

           Summary: assert building libstdc++ for i686-pc-mingw32
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: deeppatel1987 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Building with r76261.

Assertion failed: (IndexReg.getReg() != X86::ESP && "X86 doesn't allow scaling
by ESP"), function printLeaMemReference, file
/Volumes/Data/Projects/llvm-top/llvm/lib/Target/X86/AsmPrinter/X86ATTAsmPrinter.cpp,
line 495.

bugpoint reduction attached.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 17 21:27:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 17 Jul 2009 21:27:27 -0500
Subject: [LLVMbugs] [Bug 4573] New: excessive code size with X86 backend
Message-ID: <bug-4573-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4573

           Summary: excessive code size with X86 backend
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: OpenBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jsg at openbsd.org
                CC: llvmbugs at cs.uiuc.edu


When compiling various parts of the OpenBSD kernel there is a significant
increase in code size on many object files, which is likely the cause
of slow compilation speeds.

Such as:
if_spppsubr.o, clang: 124440, gcc335: 48968, gcc424: 49272, llvmgcc: 120400
bwi.o, clang: 164452, gcc335: 99856, gcc424: 101664, llvmgcc: 163172
pciide.o, clang: 90628, gcc335: 74560, gcc424: 77036, llvmgcc: 88160

A full list can be found at
http://goblin.cx/llvm/LLVM-bakeoff-comparison
and a summary at
http://goblin.cx/llvm/LLVM-bakeoff

Compile flags used
-Wall -Wstrict-prototypes -Wmissing-prototypes  -Wno-uninitialized -Wno-format
-fno-builtin -fno-zero-initialized-
in-bss  -mno-mmx -mno-sse -mno-sse2 -mno-sse3 -O2 -pipe -nostdinc

Local changes to the llvm tree are ignoring kprintf format attribute/bounded
attribute.  Changed --mcpu default from pentium4 to i486 so floating point code
isn't generated in the kernel.  And some openbsd link target changes I'll be
submitting sometime soon (mostly just libgcc).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 01:39:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 01:39:26 -0500
Subject: [LLVMbugs] [Bug 4574] New: Make inliner cost calculations reduce
 the cost for a load from a constant global
Message-ID: <bug-4574-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4574

           Summary: Make inliner cost calculations reduce the cost for a
                    load from a constant global
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


See
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081752.html;
essentially, we want to count loads from constants as constant when calculating
costs in the inliner.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 04:08:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 04:08:19 -0500
Subject: [LLVMbugs] [Bug 4214] Internal compile error when function declared
 with void formal parameter
In-Reply-To: <bug-4214-206@http.llvm.org/bugs/>
Message-ID: <200907180908.n6I98JHh009986@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4214


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #9 from Duncan Sands <baldrick at free.fr>  2009-07-18 04:08:14 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090713/081594.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 08:27:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 08:27:22 -0500
Subject: [LLVMbugs] [Bug 4575] New: llvm-gcc: user-after-free in tree
 conversion causing ICE
Message-ID: <bug-4575-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4575

           Summary: llvm-gcc: user-after-free in tree conversion causing ICE
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3210)
 --> (http://llvm.org/bugs/attachment.cgi?id=3210)
gzipped wxluxgui.i

With SVN r76312 llvm-g++ ICEs when compiling either wxluxgui.cpp or
wxluxapp.cpp
from Luxrender (www.luxrender.net):
$ /usr/local/bin/llvm-g++ wxluxgui.i  -c -emit-llvm -O0
/home/edwin/lux/n/lux-a66b768345ed/renderer/wxluxgui.cpp:3755: internal
compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

It doesn't fail always when compiling this file, but it does fail when
compiling either this one or wxluxapp.cpp (thus never being able to compile
luxrenderer).

Valgrind messages:
==9040== Invalid read of size 8                                                 
==9040==    at 0x747520: T.2842 (DenseMap.h:366)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                     
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                    
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                    
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==    by 0xEF0CBB: (within
/usr/local/libexec/gcc/x86_64-unknown-linux-gnu/4.2.1/cc1plus)                  
==9040==  Address 0x4c41268 is 120 bytes inside a block of size 1,024 free'd    
==9040==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)      
==9040==    by 0x74751F: T.2842 (DenseMap.h:469)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==                                                                        
==9040== Invalid read of size 8                                                 
==9040==    at 0x74BC2B: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1731)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==    by 0xEF0CBB: (within
/usr/local/libexec/gcc/x86_64-unknown-linux-gnu/4.2.1/cc1plus)                  
==9040==    by 0x247946BF: ???                                                  
==9040==  Address 0x4c41268 is 120 bytes inside a block of size 1,024 free'd    
==9040==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)      
==9040==    by 0x74751F: T.2842 (DenseMap.h:469)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==                                                                        
==9040== Invalid read of size 8                                                 
==9040==    at 0x74BC34: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1732)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==    by 0xEF0CBB: (within
/usr/local/libexec/gcc/x86_64-unknown-linux-gnu/4.2.1/cc1plus)                  
==9040==    by 0x247946BF: ???                                                  
==9040==  Address 0x4c41268 is 120 bytes inside a block of size 1,024 free'd    
==9040==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)      
==9040==    by 0x74751F: T.2842 (DenseMap.h:469)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==                                                                        
==9040== Invalid read of size 8                                                 
==9040==    at 0x74BC40: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1733)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==    by 0xEF0CBB: (within
/usr/local/libexec/gcc/x86_64-unknown-linux-gnu/4.2.1/cc1plus)                  
==9040==    by 0x247946BF: ???                                                  
==9040==  Address 0x4c41268 is 120 bytes inside a block of size 1,024 free'd    
==9040==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)      
==9040==    by 0x74751F: T.2842 (DenseMap.h:469)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==                                                                        
==9040== Invalid read of size 8                                                 
==9040==    at 0x74BC50: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1736)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???                                                         
==9040==    by 0xEF0CBB: (within
/usr/local/libexec/gcc/x86_64-unknown-linux-gnu/4.2.1/cc1plus)                  
==9040==    by 0x247946BF: ???                                                  
==9040==  Address 0x4c41268 is 120 bytes inside a block of size 1,024 free'd    
==9040==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)      
==9040==    by 0x74751F: T.2842 (DenseMap.h:469)                                
==9040==    by 0x74BC1F: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1728)                                 
==9040==    by 0x74C6D5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:707)                                                
==9040==    by 0x74CBCB: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:836)                                                
==9040==    by 0x74F529: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned int&, llvm::AttrListPtr&)
(llvm-types.cpp:1214)                                                           
==9040==    by 0x71D15E: make_decl_llvm (llvm-backend.cpp:1543)                 
==9040==    by 0x6FA116: assemble_alias (varasm.c:5283)                         
==9040==    by 0x485963: use_thunk (method.c:310)                               
==9040==    by 0x489FC9: expand_body (semantics.c:3189)                         
==9040==    by 0x23D8BCFF: ???                                                  
==9040==    by 0x1: ???


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 09:05:43 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 09:05:43 -0500
Subject: [LLVMbugs] [Bug 4576] New: clang-cc -ast-dump crashes
Message-ID: <bug-4576-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4576

           Summary: clang-cc -ast-dump crashes
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: AST
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: xuzhongxing at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang-cc crashes on the following code:

typedef __typeof__(sizeof(int)) size_t;
size_t f(char*);
int
ar_name (char *name)
{
  char *p;
  char *end;

  if (p == 0 || p == name)
    return 0;
  end = p + f(p) - 1;
  return 0;
}

$ clang-cc -ast-dump foo.c

typedef char *__builtin_va_list;
0   clang-cc  0x08c7acea
1   clang-cc  0x08c7b21d
2             0xb7f73400 __kernel_sigreturn + 0
3   clang-cc  0x0845d3e9
4   clang-cc  0x0824e22f
5   clang-cc  0x08469ad2
6   clang-cc  0x084679dc
7   clang-cc  0x0841f3c9
8   clang-cc  0x084229fa
9   clang-cc  0x084217bf
10  clang-cc  0x084227ae
11  clang-cc  0x08422987
12  clang-cc  0x08421272
13  clang-cc  0x08084fd1
14  clang-cc  0x082e7650
15  clang-cc  0x0806bd5e
16  clang-cc  0x0806db68 main + 3450
17  libc.so.6 0xb7cd1775 __libc_start_main + 229
18  clang-cc  0x08063c31
Stack dump:
0.      Program arguments: clang-cc -ast-dump foo.c 
1.      <eof> parser at end of file
segmentation fault


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 12:14:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 12:14:05 -0500
Subject: [LLVMbugs] [Bug 4577] New: gcc compatibility, postincrement operator
Message-ID: <bug-4577-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4577

           Summary: gcc compatibility, postincrement operator
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Basic
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sbi_ at gmx.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3211)
 --> (http://llvm.org/bugs/attachment.cgi?id=3211)
sample source

sample source 1.c (attached):
#include <stdio.h>

int main (int argc, const char** argv) {

        int a = 1;

        int b = a++ + a++;

        printf("a=%d, b=%d\n", a ,b);

        return 0;
}

compiled with gcc:
gcc --version
gcc (GCC) 4.1.2 (Gentoo 4.1.2 p1.1)
Copyright (C) 2006 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

gcc -o 1 -Wall 1.c
./1
a=3, b=2

compiled with clang:

/usr/local/clang/bin/clang --version
clang version 1.0 (http://llvm.org/svn/llvm-project/cfe/trunk 76320)
Target: i386-pc-linux-gnu
Thread model: posix

/usr/local/clang/bin/clang -o /tmp/1 1.c

/tmp/1
a=3, b=3

If I understand the C Standard correctly, in the example above b should be
assigned the result of a + a, then a should be incremented twice.  So it seems
gcc is right, clang is wrong.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 13:19:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 13:19:31 -0500
Subject: [LLVMbugs] [Bug 4577] gcc compatibility, postincrement operator
In-Reply-To: <bug-4577-206@http.llvm.org/bugs/>
Message-ID: <200907181819.n6IIJVI4028397@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4577


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-18 13:19:30 ---
a++ + a++ has undefined behavior; see http://c-faq.com/expr/evalorder2.html .


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 13:49:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 13:49:39 -0500
Subject: [LLVMbugs] [Bug 4574] Make inliner cost calculations reduce the
 cost for a load from a constant global
In-Reply-To: <bug-4574-206@http.llvm.org/bugs/>
Message-ID: <200907181849.n6IIndeh029553@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4574


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |LATER




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-18 13:49:33 ---
I just added a fixme in r76324


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 18 20:39:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 18 Jul 2009 20:39:39 -0500
Subject: [LLVMbugs] [Bug 4578] New: IfConverter::IfConvertDiamond not
 eliminating empty blocks
Message-ID: <bug-4578-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4578

           Summary: IfConverter::IfConvertDiamond not eliminating empty
                    blocks
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: deeppatel1987 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3212)
 --> (http://llvm.org/bugs/attachment.cgi?id=3212)
reduction

I reduced this case from a recent thread on "speed and code size issues." The
generated code has an unconditional branch to the next instructions and some
block entry labels.

It looks like IfConvertDiamond (possibly MergeBlocks within) aren't getting rid
of the empty true/false blocks.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 00:51:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 00:51:19 -0500
Subject: [LLVMbugs] [Bug 4580] New: warn on attribute used applied to local
	decl
Message-ID: <bug-4580-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4580

           Summary: warn on attribute used applied to local decl
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


I think that attribute used applied to a local decl is meaningless, and should
probably warn about it.
--
int f2() {
  return 0;
  {
    static int x[100] __attribute__((used));
    return x[1];
  }
}
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 12:48:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 12:48:19 -0500
Subject: [LLVMbugs] [Bug 4583] New: Assertion failed: (UFMO.isImm() && "
 Expecting flag operand on inline asm")
Message-ID: <bug-4583-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4583

           Summary: Assertion failed: (UFMO.isImm() && "Expecting flag
                    operand on inline asm")
           Product: new-bugs
           Version: unspecified
          Platform: PC
               URL: http://fxr.watson.org/fxr/source/cam/cam_periph.c
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Created an attachment (id=3213)
 --> (http://llvm.org/bugs/attachment.cgi?id=3213)
cam_periph.bc

Using clang r76368.

Assertion failed: (UFMO.isImm() && "Expecting flag operand on inline asm"),
function isRegTiedToDefOperand, file MachineInstr.cpp, line 791.
Stack dump:
0.      Program arguments: /usr/opt/llvm/bin/../libexec/clang-cc -triple
x86_64-unknown-freebsd8.0 -S -disable-free -main-file-name cam_periph.c
--relocation-model static --disable-fp-elim --unwind-tables=1
--disable-red-zone --no-implicit-float -mcpu nocona -target-feature -sse
-target-feature -sse2 -target-feature -sse3 -target-feature -mmx
-target-feature -3dnow --fmath-errno=1 -g -nostdinc -include opt_global.h
-D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -I. -I/usr/src/sys
-I/usr/src/sys/contrib/altq -O2 -Wall -Wredundant-decls -Wnested-externs
-Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual
-Wundef -Wno-pointer-sign -Werror -std=c99 -ffreestanding --stack-protector=1
-fno-common -fdiagnostics-show-option -o /tmp/cc-HrIzvI.s -x c
/usr/src/sys/cam/cam_periph.c
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'Two-Address instruction pass' on function
'@cam_periph_release'
clang: error: compiler command failed due to signal 6 (use -v to see
invocation)

bugpoint crashed when attempting to reduce the testcase, sorry.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 14:12:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 14:12:10 -0500
Subject: [LLVMbugs] [Bug 4583] Assertion failed: (UFMO.isImm() && "
 Expecting flag operand on inline asm")
In-Reply-To: <bug-4583-206@http.llvm.org/bugs/>
Message-ID: <200907191912.n6JJCA94012594@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4583


Jakob Stoklund Olesen <stoklund at 2pi.dk> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |stoklund at 2pi.dk
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Jakob Stoklund Olesen <stoklund at 2pi.dk>  2009-07-19 14:12:10 ---
Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=76373


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 14:14:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 14:14:38 -0500
Subject: [LLVMbugs] [Bug 4584] New: linkonce doesn't asmprint right on ELF
	targets
Message-ID: <bug-4584-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4584

           Summary: linkonce doesn't asmprint right on ELF targets
           Product: libraries
           Version: 1.0
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


Apparently:

define linkonce void @"foo bar"() { ret void }

prints out:

 .section .gnu.linkonce.t.foo bar,"ax", at progbits

which isn't happy.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 14:24:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 14:24:37 -0500
Subject: [LLVMbugs] [Bug 4580] warn on attribute used applied to local decl
In-Reply-To: <bug-4580-206@http.llvm.org/bugs/>
Message-ID: <200907191924.n6JJObij013206@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4580


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-19 14:24:37 ---
No, this is useful.

int f2() {
    static int x[100] __attribute__((used));
    asm(" something using x magically")
    return x[1];
  }

Another example is dropping something into a specific section, etc. clang and
llvm-gcc both get this right.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 14:29:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 14:29:42 -0500
Subject: [LLVMbugs] [Bug 4585] New: miscompilation at -O1 of
	luxrender/luxconsole
Message-ID: <bug-4585-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4585

           Summary: miscompilation at -O1 of luxrender/luxconsole
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3216)
 --> (http://llvm.org/bugs/attachment.cgi?id=3216)
4710.i

Running luxconsole built with llvm-g++ -O1 results in:
[Lux Segmentation fault

-O0 is fine. GCC 4.3 and 4.4 compile luxrender just fine.

I've reduced it to compiling just luxconsole.cpp and api.cpp:

the reduced luxconsole.cpp has (after a bunch of includes):
void foobar() {
      boost::posix_time::time_duration td(0, 0, 1, 0);
      std::stringstream ss;
      ss << td;
  }
    int main(int ac, char *av[]) {
      std::stringstream ss;
      luxError(0, 0, ss.str().c_str());
  }

If compile this at -O1, and link with api.cpp compiled at whatever -O level,
the result is an executable that segfaults. Compiling this at -O0 and api.cpp
at -O1, -O2, ... works.

If I remove the foobar function, or make it static then the segfault
dissappears.
This is strange, because foobar() is  never called/referenced (try renaming to
something random, still crashes). Also if I remove the boost:: stuff from
foobar(), the crash goes away.

Putting luxconsole.cpp and api.cpp in one C++ file, and compiling that at -O1
makes the bug go away too.

Printing something else than ss.str().c_str() (like a constant string) makes
the bug go away too.

Attached is 4710.i, which is the reduced luxconsole.cpp, to reproduce just
compile&run like this:
$ llvm-g++ 4710.i api.i -O && ./a.out
[Lux Segmentation fault

The original luxconsole was valgrind clean, this reduced one has some valgrind
warnings, but works fine with g++, it outputs:
[Lux 2009-Jul-19 22:29:142009-07-19 INFO : 0]


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 15:52:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 15:52:54 -0500
Subject: [LLVMbugs] [Bug 4587] New: Assertion failed: (Reg && "this is not a
	register!")
Message-ID: <bug-4587-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4587

           Summary: Assertion failed: (Reg && "this is not a register!")
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3220)
 --> (http://llvm.org/bugs/attachment.cgi?id=3220)
bugpoint-reduced-simplified.bc

Building emacs/lib-src/regex.c with clang/llvm r76379 on FreeBSD/amd64 results
in:

# llc bugpoint-reduced-simplified.bc
Assertion failed: (Reg && "this is not a register!"), function
isPhysicalRegister, file
/usr/src-local/llvm/include/llvm/Target/TargetRegisterInfo.h, line 309.
Stack dump:
0.      Program arguments: llc bugpoint-reduced-simplified.bc
1.      Running pass 'X86 AT&T-Style Assembly Printer' on function
'@regex_compile'
Abort (core dumped)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 16:07:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 16:07:34 -0500
Subject: [LLVMbugs] [Bug 4588] New: GVN: deletes call to _Unwind_Resume
 causing C++ programs with EH to crash
Message-ID: <bug-4588-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4588

           Summary: GVN: deletes call to _Unwind_Resume causing C++ programs
                    with EH to crash
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3221)
 --> (http://llvm.org/bugs/attachment.cgi?id=3221)
bugpoint-tooptimize.bc

I have reduced the crash in luxconsole on startup with -O1, it is GVN's fault,
it completely deletes the loads from %eh_*, thus causing a subsequent DSE pass
to delete the stores too, which causes the program to crash on startup when
this code is reached.


--- bugpoint-tooptimize.ll      2009-07-20 00:02:30.000000000 +0300    
+++ opt.ll      2009-07-20 00:03:08.000000000 +0300                    
@@ -1,4 +1,4 @@                                                        
-; ModuleID = 'bugpoint-tooptimize.bc'                                 
+; ModuleID = 'opt.bc'                                                 
 target datalayout =
"e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" 
 target triple = "x86_64-unknown-linux-gnu"                                     
        type { i64, i64 }               ; type %0                               
@@ -1386,12 +1386,12 @@                                                         

 define void @_Z6foobarv() {                                                    
 entry:                                                                         
-       %save_filt.1054 = alloca i64            ; <i64*> [#uses=2]              
-       %save_eptr.1053 = alloca i8*            ; <i8**> [#uses=2]              
+       %save_filt.1054 = alloca i64            ; <i64*> [#uses=1]              
+       %save_eptr.1053 = alloca i8*            ; <i8**> [#uses=1]              
        %ss = alloca
%"struct.std::basic_stringstream<char,std::char_traits<char>,std::allocator<char>
>"               ;
<%"struct.std::basic_stringstream<char,std::char_traits<char>,std::allocator<char>
>"*> [#uses=4]                                                               
        %td = alloca %"struct.boost::posix_time::time_duration"         ;
<%"struct.boost::posix_time::time_duration"*> [#uses=2]            
-       %eh_exception = alloca i8*              ; <i8**> [#uses=7]              
-       %eh_selector = alloca i64               ; <i64*> [#uses=4]              
+       %eh_exception = alloca i8*              ; <i8**> [#uses=3]              
+       %eh_selector = alloca i64               ; <i64*> [#uses=3]              
        %"alloca point" = bitcast i32 0 to i32          ; <i32> [#uses=0]       
        call void
@_ZN5boost10posix_time13time_durationC1Eiiil(%"struct.boost::posix_time::time_duration"*
%td, i32 0, i32 0, i32 1, i64 0)  
        %0 = call i32 @_ZStorSt13_Ios_OpenmodeS_(i32 16, i32 8) nounwind       
        ; <i32> [#uses=1]                                    
@@ -1405,55 +1405,30 @@                                                         

 invcont:               ; preds = %entry                                        
        call void
@_ZNSt18basic_stringstreamIcSt11char_traitsIcESaIcEED1Ev(%"struct.std::basic_stringstream<char,std::char_traits<char>,std::allocator<char>
>"* %ss)                                                                        
-       br label %return                                                        
-                                                                               
-bb:            ; preds = %ppad                                                 
-       %eh_select = load i64* %eh_selector             ; <i64> [#uses=1]       
-       store i64 %eh_select, i64* %save_filt.1054, align 8                     
-       %eh_value = load i8** %eh_exception             ; <i8*> [#uses=1]       
-       store i8* %eh_value, i8** %save_eptr.1053, align 8                      
-       invoke void
@_ZNSt18basic_stringstreamIcSt11char_traitsIcESaIcEED1Ev(%"struct.std::basic_stringstream<char,std::char_traits<char>,std::allocator<char>
>"* %ss)                                                                        
-                       to label %invcont1 unwind label %lpad5                  
-                                                                               
-invcont1:              ; preds = %bb                                           
-       %6 = load i8** %save_eptr.1053, align 8         ; <i8*> [#uses=1]       
-       store i8* %6, i8** %eh_exception, align 8                               
-       %7 = load i64* %save_filt.1054, align 8         ; <i64> [#uses=1]       
-       store i64 %7, i64* %eh_selector, align 8                                
-       br label %Unwind                                                        
+       ret void                                                                

-bb2:           ; preds = %ppad9                                                
-       call void @_ZSt9terminatev() noreturn nounwind                          
+invcont1:              ; preds = %lpad                                         
+       store i8* %eh_ptr, i8** %eh_exception, align 8                          
+       store i64 %eh_select4, i64* %eh_selector, align 8                       
+       call void @_Unwind_Resume(i8* %eh_ptr)                                  
        unreachable                                                             

-return:                ; preds = %invcont                                      
-       ret void                                                                
-                                                                               
 lpad:          ; preds = %entry                                                
-       %eh_ptr = call i8* @llvm.eh.exception()         ; <i8*> [#uses=1]       
+       %eh_ptr = call i8* @llvm.eh.exception()         ; <i8*> [#uses=5]       
        store i8* %eh_ptr, i8** %eh_exception                                   
-       %eh_ptr3 = load i8** %eh_exception              ; <i8*> [#uses=1]       
-       %eh_select4 = call i64 (i8*, i8*, ...)* @llvm.eh.selector.i64(i8*
%eh_ptr3, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null)     
   ; <i64> [#uses=1]                                                            
+       %eh_select4 = call i64 (i8*, i8*, ...)* @llvm.eh.selector.i64(i8*
%eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null)      
   ; <i64> [#uses=3]
        store i64 %eh_select4, i64* %eh_selector
-       br label %ppad
+       store i64 %eh_select4, i64* %save_filt.1054, align 8
+       store i8* %eh_ptr, i8** %save_eptr.1053, align 8
+       invoke void
@_ZNSt18basic_stringstreamIcSt11char_traitsIcESaIcEED1Ev(%"struct.std::basic_stringstream<char,std::char_traits<char>,std::allocator<char>
>"* %ss)
+                       to label %invcont1 unwind label %lpad5

-lpad5:         ; preds = %bb
-       %eh_ptr6 = call i8* @llvm.eh.exception()                ; <i8*>
[#uses=1]
+lpad5:         ; preds = %lpad
+       %eh_ptr6 = call i8* @llvm.eh.exception()                ; <i8*>
[#uses=2]
        store i8* %eh_ptr6, i8** %eh_exception
-       %eh_ptr7 = load i8** %eh_exception              ; <i8*> [#uses=1]
-       %eh_select8 = call i64 (i8*, i8*, ...)* @llvm.eh.selector.i64(i8*
%eh_ptr7, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1)        
   ; <i64> [#uses=1]
+       %eh_select8 = call i64 (i8*, i8*, ...)* @llvm.eh.selector.i64(i8*
%eh_ptr6, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1)        
   ; <i64> [#uses=1]
        store i64 %eh_select8, i64* %eh_selector
-       br label %ppad9
-
-ppad:          ; preds = %lpad
-       br label %bb
-
-ppad9:         ; preds = %lpad5
-       br label %bb2
-
-Unwind:                ; preds = %invcont1
-       %eh_ptr10 = load i8** %eh_exception             ; <i8*> [#uses=1]
-       call void @_Unwind_Resume(i8* %eh_ptr10)
+       call void @_ZSt9terminatev() noreturn nounwind
        unreachable
 }


To reproduce run:
opt -gvn bugpoint-tooptimize.bc, and look at the output (see diff above),
if you run -dse too, then the EH stuff is completely broken.

I have debugged this using bugpoint like so:
bugpoint -Xlinker="-Wl,/tmp/api.o,-lstdc++" -safe-run-custom -run-custom
-exec-command ./test2.sh -simplifycfg -domtree -domfrontier -mem2reg
-instcombine -simplifycfg -simplify-libcalls -instcombine -jump-threading
-simplifycfg -domtree -domfrontier -scalarrepl -instcombine -break-crit-edges
-condprop -tailcallelim -simplifycfg -reassociate -domtree -loops -loopsimplify
-domfrontier -lcssa -loop-rotate -licm -lcssa -loop-unswitch -instcombine
-scalar-evolution -lcssa -iv-users -indvars -loop-deletion -instcombine -memdep
-gvn -memdep -memcpyopt -sccp -instcombine -break-crit-edges -condprop -domtree
-memdep -dse -adce -simplifycfg -preverify -domtree -verify x.bc


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 21:05:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 21:05:52 -0500
Subject: [LLVMbugs] [Bug 4589] New: Invalid GEP optimization after bitcast
Message-ID: <bug-4589-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4589

           Summary: Invalid GEP optimization after bitcast
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: stein at ir.iit.edu
                CC: llvmbugs at cs.uiuc.edu


A bitcast followed by a getelementptr causes an
optimization in instcombine that only works on
some machines, but is applied generally.

Here is a small test program:

*================================================*
%dt = type {i32, i32 *}

define i32 **@get_el(i8 *%arg) {
  %der = bitcast i8 *%arg to %dt *
  %ptr = getelementptr %dt *%der, i32 0, i32 1
  ret i32 **%ptr
}
*================================================*

Run it through:
opt -instcombine

And you get:
*================================================*
        %dt = type { i32, i32* }

define i32** @get_el(i8* %arg) {
        %ptr = getelementptr i8* %arg, i64 8            ; <i8*> [#uses=1]
        %1 = bitcast i8* %ptr to i32**          ; <i32**> [#uses=1]
        ret i32** %1
}
*================================================*
This will happen to work on AMD64 where
dt's 2nd member's offset is 8.
However, this optimization is also applied on 32-bit systems,
where the offset should instead be 4.
So, ptr will then point to the wrong place.
Thus, the optimization is invalid.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 22:05:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 22:05:48 -0500
Subject: [LLVMbugs] [Bug 4589] Invalid GEP optimization after bitcast
In-Reply-To: <bug-4589-206@http.llvm.org/bugs/>
Message-ID: <200907200305.n6K35m9e029401@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4589


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wendling at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #1 from Bill Wendling <wendling at apple.com>  2009-07-19 22:05:45 ---
I'm assuming that the original code is this:

$ cat t.c
#include <inttypes.h>

struct dt {
  int32_t f1;
  int32_t *f2;
};

int32_t **get_el(char *arg) {
  struct dt *ptr = (struct dt*)arg;
  return &ptr->f2;
}

On a 32-bit machine, we get:

$ llvm-gcc -o - t.c -S -emit-llvm -Os
; ModuleID = 't.c'
target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
target triple = "i386-apple-darwin9.8"

define i32** @get_el(i8* %arg) nounwind readnone optsize {
entry:
        %0 = getelementptr i8* %arg, i32 4              ; <i8*> [#uses=1]
        %1 = bitcast i8* %0 to i32**            ; <i32**> [#uses=1]
        ret i32** %1
}

On a 64-bit machine, we get:

$ llvm-gcc -o - t.c -S -emit-llvm -Os -m64
; ModuleID = 't.c'
target datalayout =
"e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
target triple = "x86_64-apple-darwin9.8"

define i32** @get_el(i8* %arg) nounwind readnone optsize {
entry:
        %0 = getelementptr i8* %arg, i64 8              ; <i8*> [#uses=1]
        %1 = bitcast i8* %0 to i32**            ; <i32**> [#uses=1]
        ret i32** %1
}

Both of which are correct.

Note, that LLVM code is not 100% platform independent. Because of specifics of
languages, the LLVM IR that's generated is much different for, say, 32-bit
machines as opposed to 64-bit machines.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 22:06:30 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 22:06:30 -0500
Subject: [LLVMbugs] [Bug 4589] Invalid GEP optimization after bitcast
In-Reply-To: <bug-4589-206@http.llvm.org/bugs/>
Message-ID: <200907200306.n6K36UjW029435@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4589


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|WORKSFORME                  |INVALID




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 23:35:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 23:35:08 -0500
Subject: [LLVMbugs] [Bug 4589] Invalid GEP optimization after bitcast
In-Reply-To: <bug-4589-206@http.llvm.org/bugs/>
Message-ID: <200907200435.n6K4Z8be031898@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4589


Sterling Stuart Stein <stein at ir.iit.edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |




--- Comment #2 from Sterling Stuart Stein <stein at ir.iit.edu>  2009-07-19 23:35:07 ---
Your comment is true, but missing the point.
If you add a target field, then the optimization works correctly.
I was able to duplicate what you gave in your post.
However, the difficult arises when the
target datalayout is not given.  Apparently, without it,
the optimization assumes 64-bit.

I understand that BitCode is not entirely platform independent.
However, it should refrain from making platform-specific
optimizations when it doesn't actually know what the platform is.
When you compile that code (without the 32-bit target datalayout)
on a 32-bit machine, it still compiles it as if it was
for a 64-bit machine.  Hence, it is a bug.

Side question: What's the advantage of saying GEP 8 over GEP 0,1 anyways?
Won't LLVM generate the same code either way (but with delaying the
offset calculation until it knows which platform it is on)?
Or do other optimizations depend on this somehow?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 19 23:49:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 19 Jul 2009 23:49:39 -0500
Subject: [LLVMbugs] [Bug 4590] New: Infinite loop in ConvertToPacked in
	llvm-gcc
Message-ID: <bug-4590-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4590

           Summary: Infinite loop in ConvertToPacked in llvm-gcc
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: Nathan.Keynes at sun.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3226)
 --> (http://llvm.org/bugs/attachment.cgi?id=3226)
Test case extracted from linux drivers/usb/gadget/ether.c

Attached test case loops forever in ConvertToPacked (llvm-convert.cpp) trying
to add ever-increasing amounts of padding to the packed structure. This looks
like a regression as I can't reproduce with 2.5.

It looks like it's getting confused by the iMACAddress field, which is declared
in the structure but missing from the constant - it actually works properly if
either the iMACAddress field is removed entirely, or if an explicit value is
added for the field in the constant decl.

Possible patch:
Index: gcc/llvm-convert.cpp
===================================================================
--- gcc/llvm-convert.cpp        (revision 76305)
+++ gcc/llvm-convert.cpp        (working copy)
@@ -7368,6 +7368,7 @@
       PadTy = Context.getArrayType(PadTy, AlignedEltOffs-EltOffs);
     ResultElts.insert(ResultElts.begin()+i,
                       Context.getNullValue(PadTy));
+    EltOffs += TD.getTypeAllocSize(PadTy);
     ++e;  // One extra element to scan.
   }

This fixes the infinite loop, and produces the correct output for this testcase
and a few similar cases I've tried. I'm not 100% certain there aren't other
issues around this though.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 00:21:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 00:21:44 -0500
Subject: [LLVMbugs] [Bug 4589] Invalid GEP optimization after bitcast
In-Reply-To: <bug-4589-206@http.llvm.org/bugs/>
Message-ID: <200907200521.n6K5LiAi000835@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4589


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #4 from Chris Lattner <clattner at apple.com>  2009-07-20 00:21:43 ---
If you don't specify a target data string, llvm defaults to sparcv9 settings. 
This is suboptimal and tracked in bug 4542

*** This bug has been marked as a duplicate of bug 4542 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 01:59:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 01:59:50 -0500
Subject: [LLVMbugs] [Bug 4567] Invalid assembly produced due to unified
	syntax emission
In-Reply-To: <bug-4567-206@http.llvm.org/bugs/>
Message-ID: <200907200659.n6K6xorR003835@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4567


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Evan Cheng <evan.cheng at apple.com>  2009-07-20 01:59:50 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/081866.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 02:15:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 02:15:39 -0500
Subject: [LLVMbugs] [Bug 4588] GVN: deletes call to _Unwind_Resume causing
 C++ programs with EH to crash
In-Reply-To: <bug-4588-206@http.llvm.org/bugs/>
Message-ID: <200907200715.n6K7Fdx7004379@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4588


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #6 from T??r??k Edwin <edwintorok at gmail.com>  2009-07-20 02:15:39 ---
Ok, the problem is that some files #define NDEBUG, others don't, and if you
link a file compiled with NDEBUG that includes
boost/date_time/posix_time/posix_time.hpp, with one that includes it w/o
defining NDEBUG then bad things happen.

With gcc at -O1 everything is fine, and the result is valgrind clean, however
even with gcc at -O0 it segfaults if one files defines it, and the other one
doesn't.

Now llvm-gcc at -O1 optimized things differently and preserved the bad
behaviour.

Doesn't look like a bug in LLVM, sorry for the noise.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 02:16:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 02:16:09 -0500
Subject: [LLVMbugs] [Bug 4585] miscompilation at -O1 of luxrender/luxconsole
In-Reply-To: <bug-4585-206@http.llvm.org/bugs/>
Message-ID: <200907200716.n6K7G9E6004435@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4585


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #3 from T??r??k Edwin <edwintorok at gmail.com>  2009-07-20 02:16:09 ---
Bug is that some files are compiled with NDEBUG, others not


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 05:19:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 05:19:13 -0500
Subject: [LLVMbugs] [Bug 4591] New: uninitialized LLVMContext used by
 InstructionCombining pass . Kaleidoscope testcase crash.
Message-ID: <bug-4591-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4591

           Summary: uninitialized LLVMContext used by InstructionCombining
                    pass. Kaleidoscope testcase crash.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


Testcase:
echo "def test(x) 1+x;" | Kaleidoscope
ready> ready> Stack dump:
0.      Running pass 'Combine redundant instructions' on function '@test'
Segmenteringsfel

(gdb) bt
#0  0x08199c16 in llvm::LLVMContext::getConstantFP (this=0x0, V=@0xbfd3a004) at
/home/xerxes/llvm/lib/VMCore/LLVMContext.cpp:485
#1  0x0819b534 in llvm::LLVMContext::getNullValue (this=0x0, Ty=0x8a661c8) at
/home/xerxes/llvm/lib/VMCore/LLVMContext.cpp:46
#2  0x0819b7fa in llvm::LLVMContext::getConstantFPNegativeZero (this=0x0,
Ty=0x8a661c8) at /home/xerxes/llvm/lib/VMCore/LLVMContext.cpp:521
#3  0x082cdf68 in visitFAdd (this=0x8a8c310, I=@0x8a92d58) at
/home/xerxes/llvm/lib/Transforms/Scalar/InstructionCombining.cpp:2377
#4  0x08310a5c in visit (this=0x8a8c310, I=@0x0) at
/home/xerxes/llvm/include/llvm/Instruction.def:108
#5  0x0831412c in DoOneIteration (this=0x8a8c310, F=@0x8a92840, Iteration=0) at
/home/xerxes/llvm/lib/Transforms/Scalar/InstructionCombining.cpp:13079
#6  0x08314b6d in runOnFunction (this=0x8a8c310, F=@0x8a92840) at
/home/xerxes/llvm/lib/Transforms/Scalar/InstructionCombining.cpp:13157
#7  0x081b15c7 in llvm::FPPassManager::runOnFunction (this=0x8a8a990,
F=@0x8a92840) at /home/xerxes/llvm/lib/VMCore/PassManager.cpp:1372
#8  0x081b19da in llvm::FunctionPassManagerImpl::run (this=0x8a89360,
F=@0x8a92840) at /home/xerxes/llvm/lib/VMCore/PassManager.cpp:1324
#9  0x081b1c0f in llvm::FunctionPassManager::run (this=0xbfd3a8c4,
F=@0x8a92840) at /home/xerxes/llvm/lib/VMCore/PassManager.cpp:1254
#10 0x0812aad4 in FunctionAST::Codegen (this=0x8a92728) at
/home/xerxes/llvm/examples/Kaleidoscope/toy.cpp:989
#11 0x0812ec8f in main () at
/home/xerxes/llvm/examples/Kaleidoscope/toy.cpp:1010
(gdb) 

On frame #3 InstructionCombining.cpp:2377 the instruction combining pass calls
a static function of a uninitialised LLVMContext that eventually results into a
crash when the Context wants to dereference the LLVMContext::pImpl pointer in
frame #0.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 08:43:30 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 08:43:30 -0500
Subject: [LLVMbugs] [Bug 4592] New: ARM JIT 1+1=0
Message-ID: <bug-4592-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4592

           Summary: ARM JIT 1+1=0
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


Running Kaleidoscope on a armv5t sheevaplug using llvm svn 2.6 r76402:

ready> 1+1;
ready> Evaluated to 0.000000

ready> def test() 1+1;
ready> Read function definition:
define double @test() {
entry:
        ret double 2.000000e+00
}
ready> test();
ready> Evaluated to 0.000000
ready> 

ready> 0-2;
ready> Evaluated to 0.000000

... testing some other numbers to check the sanity of the JIT.

ready> 1;
ready> Evaluated to 1.000000

ready> def testtest() 1+2;
ready> Read function definition:
define double @testtest() {
entry:
        ret double 3.000000e+00
}

ready> testtest();
ready> Evaluated to 3.000000
ready> 11111111111;
ready> Evaluated to 11111111111.000000

seems ok.

...  OK now testing 0

ready> def zero() 0;
ready> Read function definition:
define double @zero() {
entry:
        ret double 0.000000e+00
}

ready> zero();
ready> Evaluated to 104.000992

ready> 0;
ready> Evaluated to 104.004898
ready> 0;
ready> Evaluated to 104.005875
ready> 0;
ready> Evaluated to 104.006851
ready> 0;
ready> Evaluated to 104.007828
ready> 0;
ready> Evaluated to 104.008804
ready> 0;
ready> Evaluated to 104.009781
ready> 0;
ready> [New Thread 0x4001f740 (LWP 27893)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x4001f740 (LWP 27893)]
0x41320000 in ?? ()
(gdb) bt
#0  0x41320000 in ?? ()
#1  0x000980d0 in main () at
/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr76402-r1/trunk/examples/Kaleidoscope/toy.cpp:1042


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 09:06:59 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 09:06:59 -0500
Subject: [LLVMbugs] [Bug 4593] New: ARM JIT BrainF -jit hello.bf Asserts
 while deleting putchar
Message-ID: <bug-4593-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4593

           Summary: ARM JIT BrainF -jit hello.bf Asserts while deleting
                    putchar
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3227)
 --> (http://llvm.org/bugs/attachment.cgi?id=3227)
hello world .bf for BrainF

./BrainF -jit /home/xerxes/hello.bf 
------- Running JIT -------
Hello World!
While deleting: i32 (i32)* %putchar
An asserting value handle still pointed to this value!
UNREACHABLE executed at
/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr76402-r1/trunk/lib/VMCore/Value.cpp:508!
Aborted
xerxes at debian:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr76402-r1/image/usr/examples$


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 09:52:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 09:52:09 -0500
Subject: [LLVMbugs] [Bug 4591] uninitialized LLVMContext used by
 InstructionCombining pass. Kaleidoscope testcase crash.
In-Reply-To: <bug-4591-206@http.llvm.org/bugs/>
Message-ID: <200907201452.n6KEq9ca032111@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4591


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-20 09:52:04 ---
Should be fixed with r76422.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 14:48:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 14:48:54 -0500
Subject: [LLVMbugs] [Bug 4587] subreg coalescing miscompile
In-Reply-To: <bug-4587-206@http.llvm.org/bugs/>
Message-ID: <200907201948.n6KJms2o010738@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4587


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Evan Cheng <evan.cheng at apple.com>  2009-07-20 14:48:54 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/081927.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 15:04:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 15:04:57 -0500
Subject: [LLVMbugs] [Bug 4575] llvm-gcc: user-after-free in tree conversion
	causing ICE
In-Reply-To: <bug-4575-206@http.llvm.org/bugs/>
Message-ID: <200907202004.n6KK4vbd011309@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4575


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Dale Johannesen <dalej at apple.com>  2009-07-20 15:04:57 ---
Let's try it.  Fixed here, I hope:
http://llvm.org/viewvc/llvm-project?rev=76459&view=rev
The testcase  never failed on Darwin so I can't verify this works, though.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 16:03:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 16:03:26 -0500
Subject: [LLVMbugs] [Bug 2717] significant code bloat vs. gcc at -Os
In-Reply-To: <bug-2717-206@http.llvm.org/bugs/>
Message-ID: <200907202103.n6KL3Q6m013316@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2717


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #3 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-20 16:03:24 ---
I don't think it was my changes, but it appears to work now.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 20:09:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 20:09:04 -0500
Subject: [LLVMbugs] [Bug 2831] instcombine turns defined add overflow into
 undefined GEP overflow
In-Reply-To: <bug-2831-206@http.llvm.org/bugs/>
Message-ID: <200907210109.n6L194hg022141@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2831


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #20 from Dan Gohman <gohman at apple.com>  2009-07-20 20:08:51 ---
The original testcase and the testcase in Comment #11 are now compiled
correctly. Instcombine no longer treates an arbitrary
inttoptr+arithmetic+ptrtoint as being equivalent to a GEP.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 20:14:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 20:14:41 -0500
Subject: [LLVMbugs] [Bug 1795] Missed optimization: eliminating loads around
 ptrtoint/inttoptr
In-Reply-To: <bug-1795-206@http.llvm.org/bugs/>
Message-ID: <200907210114.n6L1EfqJ022345@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1795


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gohman at apple.com
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #4 from Dan Gohman <gohman at apple.com>  2009-07-20 20:14:38 ---
getelementptr is now being handled more carefully, and it's no longer
safe to push an add below an inttoptr in general. This testcase is still 
theoretically optimizable, but it'll require a more elaborate approach.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 20 20:15:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 20 Jul 2009 20:15:35 -0500
Subject: [LLVMbugs] [Bug 4569] integer wrong code bug
In-Reply-To: <bug-4569-206@http.llvm.org/bugs/>
Message-ID: <200907210115.n6L1FZgJ022390@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4569


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Dan Gohman <gohman at apple.com>  2009-07-20 20:15:34 ---
This is fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/081963.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 21 02:12:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 21 Jul 2009 02:12:48 -0500
Subject: [LLVMbugs] [Bug 4598] New: clang: ICE on invalid C: pure virtual
	method called
Message-ID: <bug-4598-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4598

           Summary: clang: ICE on invalid C: pure virtual method called
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3229)
 --> (http://llvm.org/bugs/attachment.cgi?id=3229)
testcase-min.i

Using TOT clang I get a "pure virtual method called", and then a clang crash
when compiling the attached C code (which is invalid, and has lots of errors):

$ clang testcase-min.i                                                          
testcase-min.i:6:38: warning: declaration of 'struct cli_bc' will not be
visible outside of this function                        
     int cli_vm_execute(const struct cli_bc *bc, struct cli_bc_ctx *ctx, const
struct cli_bc_func *func, const struct ...        
                                     ^                                          
testcase-min.i:6:57: warning: declaration of 'struct cli_bc_ctx' will not be
visible outside of this function                       
     int cli_vm_execute(const struct cli_bc *bc, struct cli_bc_ctx *ctx, const
struct cli_bc_func *func, const struct ...           
                                                        ^                       
testcase-min.i:6:87: warning: declaration of 'struct cli_bc_func' will not be
visible outside of this function                      
     int cli_vm_execute(const struct cli_bc *bc, struct cli_bc_ctx *ctx, const
struct cli_bc_func *func, const struct ...           
                                                                               
      ^                                             
testcase-min.i:8:15: error: expected '}'                                        
  case 0: do {                                                                  
              ^                                                                 
testcase-min.i:8:15: error: expected '}'                                        
pure virtual method called                                                      
terminate called without an active exception                                    
0   clang-cc        0x00000000011626ff                                          
1   clang-cc        0x0000000001162aed
2   libpthread.so.0 0x0000003fe8a0e700
3   libc.so.6       0x0000003fe7e31d05 gsignal + 53
4   libc.so.6       0x0000003fe7e34dc1 abort + 385
5   libstdc++.so.6  0x00000030052c6dc5 __gnu_cxx::__verbose_terminate_handler()
+ 277
6   libstdc++.so.6  0x00000030052c51f6
7   libstdc++.so.6  0x00000030052c5223
8   libstdc++.so.6  0x00000030052c5aff
9   clang-cc        0x00000000006abc80
10  clang-cc        0x00000000007a25e8
11  clang-cc        0x00000000007a2feb
12  clang-cc        0x00000000007a3406
13  clang-cc        0x00000000007a3bb1
14  clang-cc        0x000000000076a107
15  clang-cc        0x000000000076afd5
16  clang-cc        0x000000000076b170
17  clang-cc        0x000000000076b676
18  clang-cc        0x00000000006117d3
19  clang-cc        0x000000000040d214
20  clang-cc        0x0000000000410e34 main + 3508
21  libc.so.6       0x0000003fe7e1e5a6 __libc_start_main + 230
22  clang-cc        0x000000000040af69
Stack dump:
0.      Program arguments: /usr/local/bin/../libexec/clang-cc -triple
x86_64-unknown-linux-gnu -S -disable-free -main-file-name testcase-min.i
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -fdiagnostics-show-option -o /tmp/cc-GHn0wB.s -x cpp-output
testcase-min.i
1.      <eof> parser at end of file
2.      testcase-min.i:6:138: parsing function body 'cli_vm_execute'
3.      testcase-min.i:6:138: in compound statement ('{}')
clang: error: compiler command failed due to signal 6 (use -v to see
invocation)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 21 05:21:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 21 Jul 2009 05:21:53 -0500
Subject: [LLVMbugs] [Bug 4599] New: opt -O1 introduces umax expressions in
 simple for loops
Message-ID: <bug-4599-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4599

           Summary: opt -O1 introduces umax expressions in simple for loops
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3230)
 --> (http://llvm.org/bugs/attachment.cgi?id=3230)
y.bc

$ opt -scalar-evolution -analyze y.bc
...
Loop for.cond: backedge-taken count is %ecx
Loop for.cond: max backedge-taken count is -1

If I run -O1 I get a much uglier form for the loop:

$ opt -O1 -scalar-evolution y.bc -analyze
....
Loop for.body: backedge-taken count is (-1 + (1 umax %ecx))
Loop for.body: max backedge-taken count is (-1 + (1 umax %ecx))


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 21 07:23:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 21 Jul 2009 07:23:52 -0500
Subject: [LLVMbugs] [Bug 4600] New: ReleaseNotes-2.6. html should mention
 API changes affecting backends
Message-ID: <bug-4600-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4600

           Summary: ReleaseNotes-2.6.html should mention API changes
                    affecting backends
           Product: Documentation
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: General docs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


I noticed that there were some API changes related to backends:
- using formatted_raw_ostream instead of raw_ostream
- change in how targets are registered:
   - needs creating a new Target variable
   - in the constructor the parent's TargetMachine has to be initialized
   - introduction of a targetinfo directory, that contains a call to
registertarget

Following the changes made to the C backend made this API update quite easy,
but nevertheless it should be documented.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 21 12:47:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 21 Jul 2009 12:47:32 -0500
Subject: [LLVMbugs] [Bug 4601] New: assertion failure: Don' t have operand
 info for this instruction!
Message-ID: <bug-4601-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4601

           Summary: assertion failure: Don't have operand info for this
                    instruction!
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bob.wilson at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3234)
 --> (http://llvm.org/bugs/attachment.cgi?id=3234)
testcase

ClamAV failed to build for ARM last night.  This is a bugpoint-reduced
testcase.  Compile it with:

llc -O3 -relocation-model=pic -mcpu=arm1136jf-s -f
bugpoint-reduced-simplified.bc

Assertion failed: ((DstRC || (TID.isVariadic() && IIOpNum >=
TID.getNumOperands())) && "Don't have operand info for this instruction!"),
function AddRegisterOperand, file
/Users/bwilson/local/llvm/llvm/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodesEmit.cpp,
line 259.
0   llc               0x00a6aa1b PrintStackTrace(void*) + 45
1   llc               0x00a6af8d SignalHandler(int) + 351
2   libSystem.B.dylib 0x96a7abfb _sigtramp + 43
3   libSystem.B.dylib 0xffffffff _sigtramp + 1767396399
4   libSystem.B.dylib 0x96b08ba1 raise + 26
5   libSystem.B.dylib 0x96b1ec58 abort + 93
6   libSystem.B.dylib 0x96b0b800 __pthread_markcancel + 0
7   llc               0x006be2c5
llvm::ScheduleDAGSDNodes::AddRegisterOperand(llvm::MachineInstr*,
llvm::SDValue, unsigned int, llvm::TargetInstrDesc const*,
llvm::DenseMap<llvm::SDValue, unsigned int, llvm::DenseMapInfo<llvm::SDValue>,
llvm::DenseMapInfo<unsigned int> >&) + 545
8   llc               0x006be462
llvm::ScheduleDAGSDNodes::AddOperand(llvm::MachineInstr*, llvm::SDValue,
unsigned int, llvm::TargetInstrDesc const*, llvm::DenseMap<llvm::SDValue,
unsigned int, llvm::DenseMapInfo<llvm::SDValue>, llvm::DenseMapInfo<unsigned
int> >&) + 86
9   llc               0x006c0089
llvm::ScheduleDAGSDNodes::EmitNode(llvm::SDNode*, bool, bool,
llvm::DenseMap<llvm::SDValue, unsigned int, llvm::DenseMapInfo<llvm::SDValue>,
llvm::DenseMapInfo<unsigned int> >&) + 793
10  llc               0x006c0aaf llvm::ScheduleDAGSDNodes::EmitSchedule() + 461
11  llc               0x0072aa3d llvm::SelectionDAGISel::CodeGenAndEmitDAG() +
4171
12  llc               0x0072c9e0
llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*,
llvm::ilist_iterator<llvm::Instruction>,
llvm::ilist_iterator<llvm::Instruction>) + 434
13  llc               0x0072d3a9
llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&,
llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*,
llvm::TargetInstrInfo const&) + 2255
14  llc               0x0072e163
llvm::SelectionDAGISel::runOnFunction(llvm::Function&) + 927
15  llc               0x009eb070
llvm::FPPassManager::runOnFunction(llvm::Function&) + 288
16  llc               0x009eb592
llvm::FunctionPassManagerImpl::run(llvm::Function&) + 112
17  llc               0x009eb75e
llvm::FunctionPassManager::run(llvm::Function&) + 130
18  llc               0x0002dfb7 main + 3142
19  llc               0x0002cc61 start + 53
Stack dump:
0.      Program arguments: /Users/bwilson/local/llvm/install/bin/llc -O3
-relocation-model=pic -mcpu=arm1136jf-s -f bugpoint-reduced-simplified.bc -o
Output/clamscan.llc.s 
1.      Running pass 'ARM Instruction Selection' on function '@cli_ac_addsig'
Abort trap


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 21 19:49:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 21 Jul 2009 19:49:04 -0500
Subject: [LLVMbugs] [Bug 4599] opt -O1 introduces umax expressions in simple
	for loops
In-Reply-To: <bug-4599-206@http.llvm.org/bugs/>
Message-ID: <200907220049.n6M0n48W019613@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4599


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #8 from Dan Gohman <gohman at apple.com>  2009-07-21 19:49:04 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082093.html

isLoopGuardedByCond continues to get smarter...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 22 09:27:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 22 Jul 2009 09:27:26 -0500
Subject: [LLVMbugs] [Bug 4605] New: ParseBitcodeFile() no longer works on
 files with @llvm. global_ctors
Message-ID: <bug-4605-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4605

           Summary: ParseBitcodeFile() no longer works on files with
                    @llvm.global_ctors
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: major
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: csdavec at swan.ac.uk
                CC: llvmbugs at cs.uiuc.edu


This simple program now crashes (tested with r76735):

#include <llvm/Bitcode/ReaderWriter.h>
#include <llvm/Support/MemoryBuffer.h>
#include <llvm/LLVMContext.h>


int main(int argc, char *argv[])
{
    ParseBitcodeFile(
            llvm::MemoryBuffer::getFile(argv[1]),
            *new llvm::LLVMContext());
}

The bug appears to have been introduced as part of the process of adding
LLVMContext.  Other appending globals appear to work correctly, but any bitcode
file containing @llvm.global_ctors crashes with this assert failure:

Assertion failed: (IMI != InverseMap.end() && IMI->second != Map.end() &&
IMI->second->second == CP && "InverseMap corrupt!"), function
FindExistingElement, file LLVMContextImpl.cpp, line 141.

Back trace as follows:

#0  0x283e2e87 in kill () from /lib/libc.so.7
#1  0x281d6347 in raise () from /lib/libthr.so.3
#2  0x283e19ea in abort () from /lib/libc.so.7
#3  0x283c7eb6 in __assert () from /lib/libc.so.7
#4  0x080f5a6b in ValueMap<std::vector<llvm::Constant*,
std::allocator<llvm::Constant*> >, llvm::ArrayType, llvm::ConstantArray,
true>::FindExistingElement (
    this=0x285170a0, CP=0x2855904c) at LLVMContextImpl.cpp:139
#5  0x080f60ba in ValueMap<std::vector<llvm::Constant*,
std::allocator<llvm::Constant*> >, llvm::ArrayType, llvm::ConstantArray,
true>::remove (
    this=0x285170a0, CP=0x2855904c) at LLVMContextImpl.cpp:208
#6  0x080f490b in llvm::LLVMContextImpl::erase (this=0x28509080, C=0x2855904c)
    at LLVMContextImpl.cpp:476
#7  0x080f0a46 in llvm::LLVMContext::erase (this=0x2850808c, C=0x2855904c)
    at LLVMContext.cpp:658
#8  0x080a47f6 in llvm::ConstantArray::destroyConstant (this=0x2855904c)
    at Constants.cpp:1041
#9  0x080522e3 in llvm::BitcodeReaderValueList::ResolveConstantForwardRefs (
    this=0x2851f270) at BitcodeReader.cpp:305
#10 0x08054ab6 in llvm::BitcodeReader::ParseConstants (this=0x2851f180)
    at BitcodeReader.cpp:1070
#11 0x08059130 in llvm::BitcodeReader::ParseModule (this=0x2851f180, 
    ModuleID=@0xbfbfe5f0) at BitcodeReader.cpp:1171
#12 0x0805a700 in llvm::BitcodeReader::ParseBitcode (this=0x2851f180)
    at BitcodeReader.cpp:1406
#13 0x0805a7c5 in llvm::getBitcodeModuleProvider (Buffer=0x2851c020, 
    Context=@0x2850809c, ErrMsg=0x0) at BitcodeReader.cpp:2104
#14 0x0805a85b in llvm::ParseBitcodeFile (Buffer=0x2851c020, 
    Context=@0x2850809c, ErrMsg=0x0) at BitcodeReader.cpp:2122
#15 0x0804f10d in main (argc=2, argv=0xbfbfe6e8) at ParseBitcode.cpp:10


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 22 19:25:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 22 Jul 2009 19:25:15 -0500
Subject: [LLVMbugs] [Bug 4609] New: Thumb2 cannot use register index with
 PC-relative addressing
Message-ID: <bug-4609-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4609

           Summary: Thumb2 cannot use register index with PC-relative
                    addressing
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bob.wilson at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3235)
 --> (http://llvm.org/bugs/attachment.cgi?id=3235)
testcase

This bug originates from trying to build darwin-crt3.c for ARMv7 in Thumb mode.
 Compile the attached testcase with:

llc -relocation-model=pic ldr_pc_plus_reg.bc

The compiler does not crash but the assembler complains:

ldr_pc_plus_reg.s:570:cannot use register index with PC-relative addressing --
`ldr r1,[pc,+r1]'

The problem does not occur if you don't specify -relocation-model=pic.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 01:08:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 01:08:53 -0500
Subject: [LLVMbugs] [Bug 4590] Infinite loop in ConvertToPacked in llvm-gcc
In-Reply-To: <bug-4590-206@http.llvm.org/bugs/>
Message-ID: <200907230608.n6N68rO9029345@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4590


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-23 01:08:48 ---
Nice catch!  I applied a slight variant of your patch here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082327.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082328.html

Thanks Nathan!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 11:58:43 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 11:58:43 -0500
Subject: [LLVMbugs] [Bug 4610] New: Assertion failed: (StructPacking == 1 &&
 "FIXME: What if StructPacking is not 1 here"), function getTypeAlignment,
 file CGRecordLayoutBuilder.cpp, line 290.
Message-ID: <bug-4610-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4610

           Summary: Assertion failed: (StructPacking == 1 && "FIXME: What if
                    StructPacking is not 1 here"), function
                    getTypeAlignment, file CGRecordLayoutBuilder.cpp, line
                    290.
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: benny.kra at gmail.com
                CC: llvmbugs at cs.uiuc.edu, andersca at mac.com


$ cat t.c
#pragma pack(4)
struct ref {
        struct ref *next;
} refs;

$ clang t.c
Assertion failed: (StructPacking == 1 && "FIXME: What if StructPacking is not 1
here"), function getTypeAlignment, file CGRecordLayoutBuilder.cpp, line 290.

Stack dump:
0.      Program arguments: /Users/d0k/Desktop/build/llvm/Debug/bin/clang-cc
-triple i386-apple-darwin9 -S -disable-free -main-file-name builtin-clone.i
--relocation-model pic -pic-level=1 --disable-fp-elim --unwind-tables=0
--mcpu=yonah --fmath-errno=0 -mmacosx-version-min=10.5.0
-fdiagnostics-show-option -o
/var/folders/OV/OVTXSzUoG4ui5ZOmbcD2uE+++TI/-Tmp-/cc-Pt6sUf.s -x cpp-output
builtin-clone.i 
1.      <eof> parser at end of file
clang: error: compiler command failed due to signal 6 (use -v to see
invocation)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 12:00:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 12:00:33 -0500
Subject: [LLVMbugs] [Bug 4611] New: Assertion failed:
 (BitsAvailableInLastField && " Bitfield size mismatch!"),
 function LayoutBitField, file CGRecordLayoutBuilder.cpp, line 67.
Message-ID: <bug-4611-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4611

           Summary: Assertion failed: (BitsAvailableInLastField && "Bitfield
                    size mismatch!"), function LayoutBitField, file
                    CGRecordLayoutBuilder.cpp, line 67.
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: benny.kra at gmail.com
                CC: llvmbugs at cs.uiuc.edu, andersca at mac.com


$ cat t.c
struct object_entry {
        unsigned int type:3, pack_id:16, depth:13;
} entries;

$ clang t.c
Assertion failed: (BitsAvailableInLastField && "Bitfield size mismatch!"),
function LayoutBitField, file CGRecordLayoutBuilder.cpp, line 67.

Stack dump:
0.      Program arguments: /Users/d0k/Desktop/build/llvm/Debug/bin/clang-cc
-triple i386-apple-darwin9 -S -disable-free -main-file-name fast-import.i
--relocation-model pic -pic-level=1 --disable-fp-elim --unwind-tables=0
--mcpu=yonah --fmath-errno=0 -mmacosx-version-min=10.5.0
-fdiagnostics-show-option -o
/var/folders/OV/OVTXSzUoG4ui5ZOmbcD2uE+++TI/-Tmp-/cc-3PDjPb.s -x cpp-output
fast-import.i 
1.      <eof> parser at end of file


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 12:05:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 12:05:52 -0500
Subject: [LLVMbugs] [Bug 4612] New: Assert failure: getConstant with a
 uint64_t value that doesn't fit in the type!
Message-ID: <bug-4612-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4612

           Summary: Assert failure: getConstant with a uint64_t value that
                    doesn't fit in the type!
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: docmach+llvm at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3236)
 --> (http://llvm.org/bugs/attachment.cgi?id=3236)
Code to reproduce bug

When compiling the attached code with -O3 the compilation fails.  It succeeds
if no optimization flags are used.

main.c:5:24: warning: implicitly declaring C library function 'malloc' with
type
      'void *(unsigned long)'
    int * largeArray = malloc(largeSize*sizeof(int));
                       ^
main.c:5:24: note: please include the header <stdlib.h> or explicitly provide a
      declaration for 'malloc'
clang-cc: SelectionDAG.cpp:872: llvm::SDValue
llvm::SelectionDAG::getConstant(uint64_t, llvm::MVT, bool): Assertion
`(EltVT.getSizeInBits() >= 64 || (uint64_t)((int64_t)Val >>
EltVT.getSizeInBits()) + 1 < 2) && "getConstant with a uint64_t value that
doesn't fit in the type!"' failed.
0   clang-cc        0x00000000011bdc7f
1   clang-cc        0x00000000011be069
2   libpthread.so.0 0x00007fe5e4aa5a10
3   libc.so.6       0x00007fe5e3dca205 gsignal + 53
4   libc.so.6       0x00007fe5e3dcb723 abort + 387
5   libc.so.6       0x00007fe5e3dc3229 __assert_fail + 233
6   clang-cc        0x0000000000d94e2b
7   clang-cc        0x0000000000db31bd
8   clang-cc        0x0000000000dd944c
9   clang-cc        0x0000000000dd9841
10  clang-cc        0x0000000000ddb5ef
11  clang-cc        0x00000000011529ea
12  clang-cc        0x0000000001152bf6
13  clang-cc        0x0000000001152e60
14  clang-cc        0x000000000044ba1d
15  clang-cc        0x000000000044c1c7
16  clang-cc        0x000000000063457d
17  clang-cc        0x0000000000431209
18  clang-cc        0x000000000043a1d2 main + 5506
19  libc.so.6       0x00007fe5e3db65c6 __libc_start_main + 230
20  clang-cc        0x000000000042ee29
Stack dump:
0.      Program arguments: /opt/src/llvm/Release/bin/clang-cc -triple
x86_64-unknown-linux-gnu -S -disable-free -main-file-name main.c
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -O3 -fdiagnostics-show-option -o /tmp/cc-UKxix3.s -x c main.c 
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'X86 DAG->DAG Instruction Selection' on function '@main'
clang: error: compiler command failed due to signal 6 (use -v to see
invocation)

This is with llvm r76880 and clang r76881


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 12:25:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 12:25:15 -0500
Subject: [LLVMbugs] [Bug 4610] Assertion failed: (StructPacking == 1 &&
 "FIXME: What if StructPacking is not 1 here"), function getTypeAlignment,
 file CGRecordLayoutBuilder.cpp, line 290.
In-Reply-To: <bug-4610-206@http.llvm.org/bugs/>
Message-ID: <200907231725.n6NHPFc2031075@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4610


Anders Carlsson <andersca at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Anders Carlsson <andersca at mac.com>  2009-07-23 12:25:13 ---
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090720/019472.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 13:42:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 13:42:13 -0500
Subject: [LLVMbugs] [Bug 4613] New: Member variable incorrectly overwritten
 when assigning value to other member variable
Message-ID: <bug-4613-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4613

           Summary: Member variable incorrectly overwritten when assigning
                    value to other member variable
           Product: tools
           Version: 2.5
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-g++
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: avleeuwen at gmail.com
                CC: baldrick at free.fr, llvmbugs at cs.uiuc.edu


Created an attachment (id=3237)
 --> (http://llvm.org/bugs/attachment.cgi?id=3237)
Testcase that shows incorrectly overwritten member variable

When compiling and running the attached program using llvm-g++ 2.5 on my
system, it will output "Test failed", indicating that a member variable that
was not written to by the user has been overwritten.

The member variable that is being incorrectly overwritten ('info') is preceded
by three int member variables and a struct member variable. Assigning a value
to the struct member variable using the return value of a function causes info
to be overwritten as well.

I compiled the file using
$ llvm-g++ -g -o test test.cpp

System details:
FreeBSD 7.2 / amd64 (64-bit)
llvm 2.5
Compiled using gcc 4.2.1 (system compiler)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 13:42:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 13:42:13 -0500
Subject: [LLVMbugs] [Bug 4614] New: icc 11.0 build issue: no
	__builtin_bswap32/64
Message-ID: <bug-4614-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4614

           Summary: icc 11.0 build issue: no __builtin_bswap32/64
           Product: Build scripts
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Makefiles
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: markleone at gmail.com
                CC: llvmbugs at cs.uiuc.edu


icc defines __GNUC__ but version 11.0 doesn't define __builtin_bswap32/64,
which is used in MathExtras.h.

If it's not easy to fix properly, a workaround would be appreciated.

icc (ICC) 11.0 20090318
FC10 Linux 2.6.25-14.fc9.x86_64 

Thanks,
Mark Leone


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 14:10:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 14:10:52 -0500
Subject: [LLVMbugs] [Bug 4614] icc 11.0 build issue: no __builtin_bswap32/64
In-Reply-To: <bug-4614-206@http.llvm.org/bugs/>
Message-ID: <200907231910.n6NJAqM8002335@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4614


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Duncan Sands <baldrick at free.fr>  2009-07-23 14:10:38 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082362.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 16:24:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 16:24:31 -0500
Subject: [LLVMbugs] [Bug 4611] Assertion failed: (BitsAvailableInLastField
 && " Bitfield size mismatch!"), function LayoutBitField,
 file CGRecordLayoutBuilder.cpp, line 67.
In-Reply-To: <bug-4611-206@http.llvm.org/bugs/>
Message-ID: <200907232124.n6NLOUs5006375@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4611


Anders Carlsson <andersca at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Anders Carlsson <andersca at mac.com>  2009-07-23 16:21:40 ---
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090720/019479.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 18:16:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 18:16:48 -0500
Subject: [LLVMbugs] [Bug 4609] Thumb2 cannot use register index with
 PC-relative addressing
In-Reply-To: <bug-4609-206@http.llvm.org/bugs/>
Message-ID: <200907232316.n6NNGmqi010314@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4609


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Evan Cheng <evan.cheng at apple.com>  2009-07-23 18:16:27 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082369.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 23 18:19:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 23 Jul 2009 18:19:33 -0500
Subject: [LLVMbugs] [Bug 4612] Assert failure: getConstant with a uint64_t
 value that doesn' t fit in the type!
In-Reply-To: <bug-4612-206@http.llvm.org/bugs/>
Message-ID: <200907232319.n6NNJXTG010406@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4612


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-07-23 18:19:25 ---
I fixed this crash (r76899), but we're now generating silently incorrect code
for this.  Victor is working on ripping out the malloc instruciton completely,
which is the real fix for this.  Please see PR715 to track that work.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 24 01:44:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 24 Jul 2009 01:44:17 -0500
Subject: [LLVMbugs] [Bug 4615] New: llvm.dbg. stoppoints may refer to the
 wrong compile_unit after llvm-link
Message-ID: <bug-4615-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4615

           Summary: llvm.dbg.stoppoints may refer to the wrong compile_unit
                    after llvm-link
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: deeppatel1987 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


llvm-link a bunch of .bc files together. Note that the resulting stoppoints all
refer to @llvm.dbg.compile_unit.type instead of the one appropriate to the
original file.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 24 04:40:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 24 Jul 2009 04:40:37 -0500
Subject: [LLVMbugs] [Bug 4615] llvm.dbg. stoppoints may refer to the wrong
 compile_unit after llvm-link
In-Reply-To: <bug-4615-206@http.llvm.org/bugs/>
Message-ID: <200907240940.n6O9ebHq010562@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4615


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |daniel at zuster.org
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Daniel Dunbar <daniel at zuster.org>  2009-07-24 04:39:55 ---


*** This bug has been marked as a duplicate of bug 4530 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 24 12:30:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 24 Jul 2009 12:30:09 -0500
Subject: [LLVMbugs] [Bug 4616] New: invalid error: "function declared
 'noreturn' should not return" for function pointers marked noreturn
Message-ID: <bug-4616-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4616

           Summary: invalid error: "function declared 'noreturn' should not
                    return" for function pointers marked noreturn
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: benny.kra at gmail.com
                CC: llvmbugs at cs.uiuc.edu, mrs at apple.com


$ cat t.c
void foo() __attribute__((noreturn));
void bar() __attribute__((noreturn));
void (*fptr)() __attribute__((noreturn)) = bar;

void foo() { fptr(); }

$ clang t.c
t.c:5:22: error: function declared 'noreturn' should not return
      [-Winvalid-noreturn]
void foo() { fptr(); }
                     ^
1 diagnostic generated.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 24 17:15:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 24 Jul 2009 17:15:23 -0500
Subject: [LLVMbugs] [Bug 4617] New: MSIL backend build failure under gcc 4.4
Message-ID: <bug-4617-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4617

           Summary: MSIL backend build failure under gcc 4.4
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: nicholas at mxc.ca
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


lib/Target/MSIL/MSILWriter.h contains an anonymous namespace. This header is
included by both MSILWriter.cpp and TargetInfo/MSILTargetInfo.cpp.

With GCC 4.4, TargetInfo/MSILTargetInfo.cpp generates a .o file that includes
undefined symbols for all of the class methods. Since those are never defined
in MSILTargetInfo.cpp, they can never be resolved and we get an undefined
symbol error at link time. (For reference, GCC 4.2 never generates bothers to
emit undef symbols because they're never mentioned outside the declaration.)

The minimal fix is to remove the anonymous namespace in the header, putting the
class declarations into the global space, but we can almost certainly do
something smarter.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 04:05:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 04:05:21 -0500
Subject: [LLVMbugs] [Bug 4618] New: Alias Analysis horks on file.
Message-ID: <bug-4618-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4618

           Summary: Alias Analysis horks on file.
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Global Analyses
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: wendling at apple.com
                CC: llvmbugs at cs.uiuc.edu


The attached .bc file horked (in an unrelated way) when I was trying to reduce
it with bugpoint:

[Irk:llvm] opt bugpoint-reduced-simplified.bc -loop-deletion -licm -indvars
-prune-eh -o foo.bc -f
0   opt               0x003b795a llvm::sys::SetInterruptFunction(void (*)()) +
82
1   opt               0x003b7d49 llvm::sys::SetInterruptFunction(void (*)()) +
1089
2   libSystem.B.dylib 0x91bf62bb _sigtramp + 43
3   libSystem.B.dylib 0xffffffff _sigtramp + 1849728367
4   opt               0x00157c35 llvm::PATypeHolder::operator llvm::Type*()
const + 17
5   opt               0x00157c4c llvm::Value::getType() const + 20
6   opt               0x0035c61a llvm::Value::getUnderlyingObject() + 114
7   opt               0x002020cf llvm::Value::getUnderlyingObject() const + 17
8   opt               0x00201367 llvm::createBasicAliasAnalysisPass() + 7575
9   opt               0x001fb16d llvm::AliasSet::aliasesPointer(llvm::Value
const*, unsigned int, llvm::AliasAnalysis&) const + 251
10  opt               0x001fc09d
llvm::AliasSetTracker::findAliasSetForPointer(llvm::Value const*, unsigned int)
+ 109
11  opt               0x001fc4c1
llvm::AliasSetTracker::getAliasSetForPointer(llvm::Value*, unsigned int, bool*)
+ 129
12  opt               0x001ff3c4
llvm::AliasSetTracker::addPointer(llvm::Value*, unsigned int,
llvm::AliasSet::AccessType, bool&) + 44
13  opt               0x001fc6d8
llvm::AliasSetTracker::add(llvm::AliasSetTracker const&) + 346
14  opt               0x000f9a8f llvm::createLICMPass() + 8565
15  opt               0x002224cf
llvm::LPPassManager::runOnFunction(llvm::Function&) + 987
16  opt               0x00335089
llvm::FPPassManager::runOnFunction(llvm::Function&) + 295
17  opt               0x00335228
llvm::FPPassManager::runOnModule(llvm::Module&) + 84
18  opt               0x00334c9f
llvm::MPPassManager::runOnModule(llvm::Module&) + 367
19  opt               0x00334ed0 llvm::PassManagerImpl::run(llvm::Module&) +
112
20  opt               0x00334f29 llvm::PassManager::run(llvm::Module&) + 27
21  opt               0x0000db9c main + 4634
22  opt               0x00002346 start + 54
23  opt               0x00000009 start + 18446744073709542649
Stack dump:
0.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
1.      Running pass 'Loop Pass Manager' on function '@main'
2.      Running pass 'Loop Invariant Code Motion' on basic block '%bb4'


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 05:35:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 05:35:48 -0500
Subject: [LLVMbugs] [Bug 4619] New: wrong session type
Message-ID: <bug-4619-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4619

           Summary: wrong session type
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rafael.espindola at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Compiling

target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32"
target triple = "i386-unknown-linux-gnu"
@foo = common global i32 0

produces

        .type   foo, at object
        .section        .gnu.linkonce.b.foo,"aw", at progbits
        .comm   foo,4,4         # foo

the use of @progbits causes the assembler to print

Warning: setting incorrect section type for .gnu.linkonce.b.foo

This was introduced by revision 76971


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 12:58:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 12:58:02 -0500
Subject: [LLVMbugs] [Bug 4619] wrong session type
In-Reply-To: <bug-4619-206@http.llvm.org/bugs/>
Message-ID: <200907251758.n6PHw26s014618@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4619


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-25 12:57:59 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090720/082543.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 14:21:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 14:21:14 -0500
Subject: [LLVMbugs] [Bug 4622] New: Assertion failed: (CurPtr <= End && "
 Unable to allocate memory!")
Message-ID: <bug-4622-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4622

           Summary: Assertion failed: (CurPtr <= End && "Unable to allocate
                    memory!")
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


This happens during the "make depend" stage of the FreeBSD kernel build.

Assertion failed: (CurPtr <= End && "Unable to allocate memory!"), function
Allocate, file Allocator.cpp, line 116.
Stack dump:
0.      Program arguments: /usr/opt/llvm/bin/../libexec/clang-cc -triple
x86_64-unknown-freebsd8.0 -Eonly -disable-free -main-file-name vm_glue.c
--relocation-model static --disable-fp-elim --unwind-tables=1
--disable-red-zone --no-implicit-float -mcpu nocona -target-feature -sse
-target-feature -sse2 -target-feature -sse3 -target-feature -mmx
-target-feature -3dnow --fmath-errno=1 -g -nostdinc -dependency-file - -MT
vm_glue.o -sys-header-deps -include opt_global.h -D_KERNEL
-DHAVE_KERNEL_OPTION_HEADERS -I. -I/usr/src/sys -I/usr/src/sys/contrib/altq
-I/usr/src/sys/contrib/ipfilter -I/usr/src/sys/contrib/pf
-I/usr/src/sys/dev/ath -I/usr/src/sys/dev/ath/ath_hal
-I/usr/src/sys/contrib/ngatm -I/usr/src/sys/dev/twa
-I/usr/src/sys/gnu/fs/xfs/FreeBSD -I/usr/src/sys/gnu/fs/xfs/FreeBSD/support
-I/usr/src/sys/gnu/fs/xfs -I/usr/src/sys/contrib/opensolaris/compat
-I/usr/src/sys/dev/cxgb -O2 -std=c99 -ffreestanding --stack-protector=1
-fno-common -fdiagnostics-show-option -x c /usr/src/sys/vm/vm_glue.c 
Abort (core dumped)

(gdb) bt
#0  0x0000000801bda65c in kill () from /lib/libc.so.7
#1  0x0000000801bd9670 in abort () from /lib/libc.so.7
#2  0x0000000801bc1a3d in __assert () from /lib/libc.so.7
#3  0x00000000011cc44e in llvm::object_creator<llvm::DenseMap<llvm::Value*,
llvm::ValueHandleBase*, llvm::DenseMapInfo<llvm::Value*>,
llvm::DenseMapInfo<llvm::ValueHandleBase*> > > ()
#4  0x000000000091a57c in clang::Parser::ParseDirectNewDeclarator ()
#5  0x000000000091c5c5 in clang::Parser::ParseDirectNewDeclarator ()
#6  0x000000000091e8ca in clang::Parser::ParseDirectNewDeclarator ()
#7  0x00000000008f6e8a in clang::Parser::ParseDirectNewDeclarator ()
#8  0x00000000008f80cb in clang::Parser::ParseDirectNewDeclarator ()
#9  0x0000000000902216 in clang::Parser::ParseDirectNewDeclarator ()
#10 0x0000000000425760 in InitializeIncludePaths ()
#11 0x00000000004284e5 in main ()


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 17:33:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 17:33:56 -0500
Subject: [LLVMbugs] [Bug 3575] warn when adding attributes following
	definition
In-Reply-To: <bug-3575-206@http.llvm.org/bugs/>
Message-ID: <200907252233.n6PMXuOt022356@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3575


Ryan Flynn <parseerror at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |parseerror at gmail.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Ryan Flynn <parseerror at gmail.com>  2009-07-25 17:33:56 ---
addressed in r77095, applies to any variable or function attribute appearing
after a definition.

$ echo "void foo(){} void foo() __attribute__((constructor));" | clang-cc
-fsyntax-only
<stdin>:1:19: warning: attribute declaration must precede definition
void foo(){} void foo() __attribute__((constructor));
                  ^
<stdin>:1:6: note: previous definition is here
void foo(){} void foo() __attribute__((constructor));
     ^
2 diagnostics generated.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat Jul 25 18:22:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 25 Jul 2009 18:22:08 -0500
Subject: [LLVMbugs] [Bug 4622] Assertion failed: (CurPtr <= End && "Unable
 to allocate memory!" )
In-Reply-To: <bug-4622-206@http.llvm.org/bugs/>
Message-ID: <200907252322.n6PNM8No023864@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4622


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-07-25 18:22:07 ---
Fixed


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 03:18:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 03:18:26 -0500
Subject: [LLVMbugs] [Bug 4617] MSIL backend build failure under gcc 4.4
In-Reply-To: <bug-4617-206@http.llvm.org/bugs/>
Message-ID: <200907260818.n6Q8IQOn014097@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4617


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Nick Lewycky <nicholas at mxc.ca>  2009-07-26 03:18:14 ---
This was fixed by Daniel Dunbar, then I modified that fix in r77146.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 09:39:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 09:39:20 -0500
Subject: [LLVMbugs] [Bug 4625] New: cpp backend does not support new API
 (LLVMContext, etc)
Message-ID: <bug-4625-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4625

           Summary: cpp backend does not support new API (LLVMContext, etc)
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: extremal.ik at gmail.com
                CC: llvmbugs at cs.uiuc.edu


; ModuleID = 'test.bc'

define void @test(i32 %value) nounwind  {
entry:
    %tmp0 = icmp eq i32 %value, 0
    ret void
}

after llvm-as and llc -march=cpp calling we have this C++ code:

// Generated by llvm2cpp - DO NOT MODIFY!


Module* makeXorEncryptModule(Module *mod) {

mod->setModuleIdentifier("test.ll.bc");

// Type Definitions
std::vector<const Type*>FuncTy_0_args;
FuncTy_0_args.push_back(IntegerType::get(32));
FunctionType* FuncTy_0 = FunctionType::get(
  /*Result=*/Type::VoidTy,
  /*Params=*/FuncTy_0_args,
  /*isVarArg=*/false);


// Function Declarations

Function* func_test = Function::Create(
  /*Type=*/FuncTy_0,
  /*Linkage=*/GlobalValue::ExternalLinkage,
  /*Name=*/"test", mod); 
func_test->setCallingConv(CallingConv::C);
AttrListPtr func_test_PAL;
{
  SmallVector<AttributeWithIndex, 4> Attrs;
  AttributeWithIndex PAWI;
  PAWI.Index = 4294967295U; PAWI.Attrs = 0  | Attribute::NoUnwind;
  Attrs.push_back(PAWI);
  func_test_PAL = AttrListPtr::get(Attrs.begin(), Attrs.end());

}
func_test->setAttributes(func_test_PAL);

// Global Variable Declarations


// Constant Definitions
ConstantInt* const_int32_1 = ConstantInt::get(APInt(32,  "0", 1, 10));

// Global Variable Definitions

// Function Definitions

// Function: test (func_test)
{
  Function::arg_iterator args = func_test->arg_begin();
  Value* int32_value = args++;
  int32_value->setName("value");

  BasicBlock* label_entry = BasicBlock::Create("entry",func_test,0);

  // Block entry (label_entry)
  ICmpInst* int1_tmp0 = new ICmpInst(ICmpInst::ICMP_EQ, int32_value,
const_int32_1, "tmp0", label_entry);
  ReturnInst::Create(label_entry);

}

return mod;

}

This code valid for llvm 2.5, but not for top-of-tree(for example
ConstantInt::get call at line 39 is obsolete).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 09:51:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 09:51:29 -0500
Subject: [LLVMbugs] [Bug 4626] New: UNREACHABLE executed at
	BranchFolding.cpp:465!
Message-ID: <bug-4626-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4626

           Summary: UNREACHABLE executed at BranchFolding.cpp:465!
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen using Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp178$ clang -O2 small.c
Predecessor appears twice
UNREACHABLE executed at BranchFolding.cpp:465!
0   clang-cc 0x091d6868
Stack dump:
0.      Program arguments:
/home/regehr/z/tmp/llvm-gcc-r77145-install/bin/../libexec/clang-cc -triple
i386-pc-linux-gnu -S -disable-free -main-file-name small.c --relocation-model
static --disable-fp-elim --unwind-tables=0 --mcpu=pentium4 --fmath-errno=1 -O2
-fdiagnostics-show-option -o /tmp/cc-yKUjit.s -x c small.c 
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'Control Flow Optimizer' on function '@foo'
clang: error: compiler command failed due to signal 6 (use -v to see
invocation)

regehr at john-home:~/volatile/tmp178$ clang -v

clang version 1.0 (http://llvm.org/svn/llvm-project/cfe/trunk 77145)
Target: i386-pc-linux-gnu
Thread model: posix

regehr at john-home:~/volatile/tmp178$ cat small.c

char safe_mul_func_int16_t_s_s (int _si1, char _si2)
{
  return _si1 && _si2 <= 0 && _si2 < _si1 || _si1 <= 0 && _si2 > 0
    && _si2 ? : _si2;
}

char g_3; 

int func_34 (char p_35)
{
  if (p_35 || g_3) return g_3;
  return 0;
}

int safe (void);

void foo (int p_5)
{
  {
    char l;
    func_34 (safe_mul_func_int16_t_s_s (1 > p_5 ^ 1, safe ()));
    for (l; 1; l = 0)
      {
      }
  }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 12:32:03 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 12:32:03 -0500
Subject: [LLVMbugs] [Bug 4628] New: clang doesn't support '-mcmodel=kernel'
Message-ID: <bug-4628-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4628

           Summary: clang doesn't support '-mcmodel=kernel'
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The FreeBSD kernel needs support for -mcmodel=kernel as of recent clang/llvm
ToT.

Currently linking dies with:
linking kernel.debug
cam.o(.text+0x62e): In function `cam_error_string':
/usr/src/sys/cam/cam.c:265: relocation truncated to fit: R_X86_64_32
.rodata.str1.1
cam.o(.text+0x6a8):/usr/src/sys/cam/cam.c:288: relocation truncated to fit:
R_X86_64_32 .rodata.str1.1
cam.o(.text+0x6d2):/usr/src/sys/cam/cam.c:291: relocation truncated to fit:
R_X86_64_32 .rodata.str1.1
cam.o(.text+0x755):/usr/src/sys/cam/cam.c:309: relocation truncated to fit:
R_X86_64_32 .rodata.str1.1
cam.o(.text+0x890): In function `cam_error_print':
/usr/src/sys/cam/cam.c:344: relocation truncated to fit: R_X86_64_32
.rodata.str1.1
...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 13:15:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 13:15:34 -0500
Subject: [LLVMbugs] [Bug 4629] New: quest test case generator finds possible
 bugs in clang and llvmc on x86-64
Message-ID: <bug-4629-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4629

           Summary: quest test case generator finds possible bugs in clang
                    and llvmc on x86-64
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Keywords: compile-fail, miscompilation
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: lee_thomas at lthomas.net
                CC: llvmbugs at cs.uiuc.edu


I had a few idle cycles on my cluster and was playing around and ran llvmc and
clang on the output of quest - the test-case generator. I used quest rev. 21,
with no special input or options. Before I realized I should stop it I had over
10000 failed test cases, 100 or so of which I've attached. I have no experience
in compilers and have no idea whether these are bugs in llvm, bugs in quest,
useless, or what... But I figured I would attach a few to let you guys know I
have a heck of a lot more if you want them. 
        I don't have time to sort out exactly where these go wrong, but see
llvm_tester_log.txt in the attached for the concatenated log of my script
culled for the cases I've attached... It indicates what happened (clang
crashed, llvmc crashed, produced code which crashed, etc). I ran both clang and
llvmc with no optimization or other flags... just e.g. clang input.c -ooutput. 
        All of the ones I've listed here fail on one of llvm 2.5 Release build
(from FreeBSD Ports), or clang 1.0 (from the same port), or llvm trunk rev.
77158 Debug build, on both FreeBSD 7.2-STABLE amd64. llvm_tester_log indicates
which, and see versions.txt in the attached for the --version output of each. I
also ran most of these on the llvmc package on Ubuntu 9.04 (kernel 2.6.28-13)
x86-64 on my laptop as a quick check... the vast majority failed as well,
though there were slight differences (e.g. producing bad code instead of
failing to compile, or vice-versa.)
        Let me know if you want the rest of the test set, or if you have
another set you'd like run... Never know when I might be able to squeeze
something in between jobs.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 14:42:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 14:42:28 -0500
Subject: [LLVMbugs] [Bug 4600] ReleaseNotes-2.6. html should mention API
 changes affecting backends
In-Reply-To: <bug-4600-206@http.llvm.org/bugs/>
Message-ID: <200907261942.n6QJgS0v027578@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4600


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Daniel Dunbar <daniel at zuster.org>  2009-07-26 14:42:27 ---
I did some of this. The LLVM backend docs are incomplete, but this was a
preexisting condition. Please put other 2.6 release note documentation requests
in the release notes as FIXMEs.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun Jul 26 20:06:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 26 Jul 2009 20:06:55 -0500
Subject: [LLVMbugs] [Bug 4616] invalid error: "function declared 'noreturn'
 should not return" for function pointers marked noreturn
In-Reply-To: <bug-4616-206@http.llvm.org/bugs/>
Message-ID: <200907270106.n6R16td1004546@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4616


Mike Stump <mrs at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|mrs at apple.com               |
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Mike Stump <mrs at apple.com>  2009-07-26 20:06:54 ---
Ok, this is fixed now.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 27 15:00:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 27 Jul 2009 15:00:17 -0500
Subject: [LLVMbugs] [Bug 3660] Clang does not enforce the restrictions on
	main
In-Reply-To: <bug-3660-206@http.llvm.org/bugs/>
Message-ID: <200907272000.n6RK0H7s019416@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3660


Douglas Gregor <dgregor at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #6 from Douglas Gregor <dgregor at apple.com>  2009-07-27 15:00:17 ---
Fixed by John McCall in:

 
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090720/019551.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 27 17:46:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 27 Jul 2009 17:46:42 -0500
Subject: [LLVMbugs] [Bug 4634] New: code gen issue for array arithmetic with
	casts
Message-ID: <bug-4634-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4634

           Summary: code gen issue for array arithmetic with casts
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: robert at muth.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3248)
 --> (http://llvm.org/bugs/attachment.cgi?id=3248)
source code for reproducer

The attached code is a reproducer for this bug.

In essence, the following macro returns a constant value:

#define bin_at(i)      ((mbinptr)((char*)&(Xav[2*(i) + 2]) - 2*sizeof(int)))


no matter what the value of i.


Disclaimer:

The problem occurred on an svn revision a couple of weeks old
on a linux cross-compilation environment.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 27 17:58:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 27 Jul 2009 17:58:22 -0500
Subject: [LLVMbugs] [Bug 4635] New: various torture test failures from the
 gcc suit ( mostly vararg stuff) on ARM
Message-ID: <bug-4635-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4635

           Summary: various torture test failures from the gcc suit (mostly
                    vararg stuff)  on ARM
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: robert at muth.org
                CC: llvmbugs at cs.uiuc.edu


The enclosed tarball contains failing tests from the gnu torture suite.
A few test do not even compile with arm backend most others trigger
problems with stdarg and vararg.

Thess bugs where found on a linux cross compilation enviroment
using an llc sync'ed on 2009/22/09


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 28 02:19:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 28 Jul 2009 02:19:48 -0500
Subject: [LLVMbugs] [Bug 4637] New: Some FP optimization opportunities missed
Message-ID: <bug-4637-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4637

           Summary: Some FP optimization opportunities missed
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


The following portion of code from Blender's fluid simulation takes up ~20-25%
of fluid simulation time (with both gcc/llvm). It looks like we are missing
some optimization opportunities here:

void bar(float *y);
void foo(float rho, float uy, float ux, float usqr, float uz) {
float lcsmeq[19];
 lcsmeq[1 ] = ( (1.0/18.0)*(rho + uy*(4.5*uy + 3.0) - usqr)) ;
 lcsmeq[2 ] = ( (1.0/18.0)*(rho + uy*(4.5*uy - 3.0) - usqr)) ;
 lcsmeq[3 ] = ( (1.0/18.0)*(rho + ux*(4.5*ux + 3.0) - usqr)) ;
 lcsmeq[4 ] = ( (1.0/18.0)*(rho + ux*(4.5*ux - 3.0) - usqr)) ;
 lcsmeq[5 ] = ( (1.0/18.0)*(rho + uz*(4.5*uz + 3.0) - usqr)) ;
 lcsmeq[6 ] = ( (1.0/18.0)*(rho + uz*(4.5*uz - 3.0) - usqr)) ;
 lcsmeq[7] = ( (1.0/36.0)*(rho + (+ux+uy)*(4.5*(+ux+uy) + 3.0) - usqr));
 lcsmeq[8] = ( (1.0/36.0)*(rho + (-ux+uy)*(4.5*(-ux+uy) + 3.0) - usqr));
 lcsmeq[9] = ( (1.0/36.0)*(rho + (+ux-uy)*(4.5*(+ux-uy) + 3.0) - usqr));
 lcsmeq[10] = ( (1.0/36.0)*(rho + (-ux-uy)*(4.5*(-ux-uy) + 3.0) - usqr));
 lcsmeq[11] = ( (1.0/36.0)*(rho + (+uy+uz)*(4.5*(+uy+uz) + 3.0) - usqr));
 lcsmeq[12] = ( (1.0/36.0)*(rho + (+uy-uz)*(4.5*(+uy-uz) + 3.0) - usqr));
 lcsmeq[13] = ( (1.0/36.0)*(rho + (-uy+uz)*(4.5*(-uy+uz) + 3.0) - usqr));
 lcsmeq[14] = ( (1.0/36.0)*(rho + (-uy-uz)*(4.5*(-uy-uz) + 3.0) - usqr));
 lcsmeq[15] = ( (1.0/36.0)*(rho + (+ux+uz)*(4.5*(+ux+uz) + 3.0) - usqr));
 lcsmeq[16] = ( (1.0/36.0)*(rho + (+ux-uz)*(4.5*(+ux-uz) + 3.0) - usqr));
 lcsmeq[17] = ( (1.0/36.0)*(rho + (-ux+uz)*(4.5*(-ux+uz) + 3.0) - usqr));
 lcsmeq[18] = ( (1.0/36.0)*(rho + (-ux-uz)*(4.5*(-ux-uz) + 3.0) - usqr));
 bar(lcsmeq);
}

Attached is the .bc file we produce, and the assembly has these:
cvtss2sd        %xmm1, %xmm5                                 
        movsd   .LCPI1_0, %xmm6                                      
        movapd  %xmm5, %xmm7                                         
        mulsd   %xmm6, %xmm7                                         
        movsd   .LCPI1_1, %xmm8                                      
        movapd  %xmm7, %xmm9                                         
        addsd   %xmm8, %xmm9                                         
        mulsd   %xmm5, %xmm9                                         
        cvtss2sd        %xmm0, %xmm0                                 
        addsd   %xmm0, %xmm9                                         
        cvtss2sd        %xmm3, %xmm3                                 
        subsd   %xmm3, %xmm9                                         
        movsd   .LCPI1_2, %xmm10                                     
        mulsd   %xmm10, %xmm9                                        
        cvtsd2ss        %xmm9, %xmm9                                 
        movss   %xmm9, 16(%rsp)                                      
        movsd   .LCPI1_3, %xmm9                                      
        addsd   %xmm9, %xmm7                                         
        mulsd   %xmm5, %xmm7                                         
        addsd   %xmm0, %xmm7                                         
        subsd   %xmm3, %xmm7                                         
        mulsd   %xmm10, %xmm7          
...

I see several opportunities to optimize here:

1) rho-usqr is computed each time:

        %13 = fadd double %0, %12               ; <double> [#uses=1]
        %14 = fsub double %13, %6               ; <double> [#uses=1]
        %15 = fmul double %14, 0x3FAC71C71C71C71C               ; <double>
[#uses=1]
        %16 = fptrunc double %15 to float               ; <float> [#uses=1]

        %22 = fadd double %0, %21               ; <double> [#uses=1]
        %23 = fsub double %22, %6               ; <double> [#uses=1]
        %24 = fmul double %23, 0x3FAC71C71C71C71C               ; <double>
[#uses=1]
        %25 = fptrunc double %24 to float       

Instead a temporary could be introduced that stores the result of (rho-usqr).
FADD/FSUB isn't associative, so this transform is not safe in general (unless
maybe there is unsafe-fpmath?), but in this case the result is truncated to a
float, so I think the following is sufficient to guarantee same results:

Calculate the rounding error introduced by applying associativity to fadd/fsub,
it can only be the last bit of the mantissa, now apply any operations that are
done on the result (in this case multiply by 1/18), and convert to float. If
the result is zero, then we can apply associativity since it only changes bits
of the double's mantissa that get truncated anyway.

2) There are lots of floating point extensions of the fsub:
        %60 = fsub float %uy, %ux               ; <float> [#uses=1]
        %61 = fpext float %60 to double         ; <double> [#uses=2]

But uy and uy are already available in extended form:
        %1 = fpext float %uy to double          ; <double> [#uses=3]
        %18 = fpext float %ux to double         ; <double> [#uses=3]

%61 could be calculated as:
%61 = fsub double %1, %18

Or would that violate some IEEE FP rules?

3) This sort of code is a very good candidate for vectorization, since it is
the exact same operations applied to different operands

If any of my above optimizations are still unsafe, then maybe we should do them
at least for -ffast-math, or -ffinite-math-only -fno-trapping-math
-fno-signaling-nans


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 28 08:37:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 28 Jul 2009 08:37:49 -0500
Subject: [LLVMbugs] [Bug 4638] New: Missing simplification of calls
Message-ID: <bug-4638-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4638

           Summary: Missing simplification of calls
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: alenhar2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


calling convention code output by the frontend is unnecessary in this case
since the call is internal and should be cleaned up to use less IR (I'm
assuming the unconstrained calling convention on i386 can efficiently handle a
pair of i32).

In the example, BlueRule should just directly return a struct.BlueReturn or
first class struct rather than allocating stackspace, storing, casting, and
loading to get an i64.  (with the matching proccess on the callee to extract
the pair).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 28 12:27:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 28 Jul 2009 12:27:15 -0500
Subject: [LLVMbugs] [Bug 4639] New: section .tbss triggers assembler warning
Message-ID: <bug-4639-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4639

           Summary: section .tbss triggers assembler warning
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


Given x.cc:

static __thread int i;
int* foo() { return &i; }

$ llvm-gcc -O2 x.cc -c -o x.o
/tmp/cc8Qh1Nc.s: Assembler messages:
/tmp/cc8Qh1Nc.s:16: Warning: setting incorrect section type for .tbss

The problem is the .section declaration:

        .section                .tbss,"awT", at progbits

which under GCC 4.2 reads:

        .section        .tbss,"awT", at nobits


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 28 12:58:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 28 Jul 2009 12:58:02 -0500
Subject: [LLVMbugs] [Bug 4639] section .tbss triggers assembler warning
In-Reply-To: <bug-4639-206@http.llvm.org/bugs/>
Message-ID: <200907281758.n6SHw24W007014@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4639


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-07-28 12:58:02 ---
Fixed in r77336, thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 28 14:28:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 28 Jul 2009 14:28:46 -0500
Subject: [LLVMbugs] [Bug 4641] New: Assertion failed: (use_empty() && " Uses
 remain when a value is destroyed!")
Message-ID: <bug-4641-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4641

           Summary: Assertion failed: (use_empty() && "Uses remain when a
                    value is destroyed!")
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3253)
 --> (http://llvm.org/bugs/attachment.cgi?id=3253)
bugpoint-reduced-simplified.bc

>From emacs lib-src/etags.c

$ opt bugpoint-reduced-simplified.bc -simplify-libcalls
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

While deleting: i8* %cond392
Use still stuck around after Def is destroyed:  %call393 = call
%struct.__sFILE* @fopen(i8* undef, i8* %cond392) nounwind               ;
<%struct.__sFILE*> [#uses=0]
Assertion failed: (use_empty() && "Uses remain when a value is destroyed!"),
function ~Value, file Value.cpp, line 78.
Stack dump:
0.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
1.      Running pass 'Simplify well-known library calls' on function '@main'
Abort (core dumped)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 00:17:59 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 00:17:59 -0500
Subject: [LLVMbugs] [Bug 4641] Assertion failed: (use_empty() && " Uses
 remain when a value is destroyed!")
In-Reply-To: <bug-4641-206@http.llvm.org/bugs/>
Message-ID: <200907290517.n6T5Hxme029636@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4641


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Nick Lewycky <nicholas at mxc.ca>  2009-07-29 00:17:59 ---
Fixed in r77433.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 00:21:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 00:21:16 -0500
Subject: [LLVMbugs] [Bug 4584] linkonce doesn't asmprint right on ELF targets
In-Reply-To: <bug-4584-206@http.llvm.org/bugs/>
Message-ID: <200907290521.n6T5LGXn029797@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4584


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Chris Lattner <clattner at apple.com>  2009-07-29 00:21:15 ---
fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090727/082852.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 05:00:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 05:00:21 -0500
Subject: [LLVMbugs] [Bug 4643] New: pathological inlining/unrolling causes
 SelectionDAG:: ComputeMaskedBits to eat time
Message-ID: <bug-4643-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4643

           Summary: pathological inlining/unrolling causes
                    SelectionDAG::ComputeMaskedBits to eat time
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: deeppatel1987 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3254)
 --> (http://llvm.org/bugs/attachment.cgi?id=3254)
crc from EEMBC

The attached test case, reduced from the EEMBC benchmark takes an excessive
amount of time to compile.

A sample report shows that we're recursing in ComputeMaskedBits a _lot_:

Total number in stack (recursive counted multiple, when >=5):
70830       llvm::SelectionDAG::ComputeMaskedBits(llvm::SDValue, llvm::APInt
const&, llvm::APInt&, llvm::APInt&, unsigned int) const
26458      
llvm::ARMTargetLowering::computeMaskedBitsForTargetNode(llvm::SDValue,
llvm::APInt const&, llvm::APInt&, llvm::APInt&, llvm::SelectionDAG const&,
unsigned int) const
371       llvm::APInt::APInt(unsigned int, unsigned long long, bool)
349       llvm::APInt::clearUnusedBits()
334       llvm::APInt::operator&(llvm::APInt const&) const
331       llvm::APInt::operator=(llvm::APInt const&)
237       llvm::APInt::isSingleWord() const
115       llvm::cast_retty<llvm::ConstantSDNode, llvm::SDValue>::ret_type
llvm::cast<llvm::ConstantSDNode, llvm::SDValue>(llvm::SDValue const&)
...

Unrolling and inlining cause some serious expansion here.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 12:31:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 12:31:14 -0500
Subject: [LLVMbugs] [Bug 4645] New: crash in simplify libcalls
Message-ID: <bug-4645-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4645

           Summary: crash in simplify libcalls
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu, nlewycky at google.com


ddunbar at lordcrumb:t$ opt -simplify-libcalls -f -o /dev/null t.bc
PHINode should have one entry for each predecessor of its parent basic block!
        %j.0 = phi i32 [ %inc47, %for.end ], [ 0, %lor.lhs.false ]             
; <i32> [#uses=1]Broken module found, compilation aborted!
0   opt               0x00000001003c4389 PrintStackTrace(void*) + 38
1   opt               0x00000001003c4899 SignalHandler(int) + 288
2   libSystem.B.dylib 0x00007fff82aa42fa _sigtramp + 26
3   libSystem.B.dylib 0x0000000101305b98 _sigtramp + 2122717368
4   opt               0x00000001003782a8 (anonymous
namespace)::Verifier::abortIfBroken() + 182
5   opt               0x0000000100385a1a (anonymous
namespace)::Verifier::runOnFunction(llvm::Function&) + 118
6   opt               0x0000000100350cfd
llvm::FPPassManager::runOnFunction(llvm::Function&) + 341
7   opt               0x0000000100350ede
llvm::FPPassManager::runOnModule(llvm::Module&) + 82
8   opt               0x000000010034ee5c
llvm::MPPassManager::runOnModule(llvm::Module&) + 388
9   opt               0x00000001003511d9
llvm::PassManagerImpl::run(llvm::Module&) + 111
10  opt               0x000000010035123b llvm::PassManager::run(llvm::Module&)
+ 33
11  opt               0x0000000100059055 main + 4435
12  opt               0x000000010004ca98 start + 52
Stack dump:
0.      Running pass 'Function Pass Manager' on module 't.bc'.
1.      Running pass 'Module Verifier' on function '@main'
Abort trap


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 13:51:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 13:51:35 -0500
Subject: [LLVMbugs] [Bug 4646] New: Key manager, EH frames,
 JIT and Mac OS X 10.5.7 (llvm 2.5)
Message-ID: <bug-4646-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4646

           Summary: Key manager, EH frames, JIT and Mac OS X 10.5.7 (llvm
                    2.5)
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: basu.samit at gmail.com
                CC: llvmbugs at cs.uiuc.edu


There seems to be an issue in LLVM v2.5 with proper operation of exceptions
within an application once the JIT is created (I know, I know, the JIT does not
do exceptions yet).  My app crashes each time it throws an exception after it
first initializes the JIT.  I noticed that in JIT.cpp in the trunk that the key
manager is not used to register EH frames (see r69090).  This patch appears to
be active for Mac OS X 10.5 and beyond -- the comment in the log suggests 10.6,
but I encounter the issue on 10.5.7.  I was unable to make HowToUseJIT
reproduce the problem, but the existence of r69090 suggests this is a known
issue.  Disabling the key manager call appears to make everything work as
expected.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 16:08:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 16:08:41 -0500
Subject: [LLVMbugs] [Bug 4648] New: JIT depends on AsmParser due to
	dwarfwriter
Message-ID: <bug-4648-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4648

           Summary: JIT depends on AsmParser due to dwarfwriter
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: dpatel at apple.com, llvmbugs at cs.uiuc.edu,
                    wendling at apple.com


Created an attachment (id=3259)
 --> (http://llvm.org/bugs/attachment.cgi?id=3259)
~/llvm-git/llvm/utils/GenLibDeps.pl -flat -why ../../Release/lib "nm -p" >out

lli is 9.7M on x86-64, some libs are linked in that shouldn't, such as
AsmParser.

llvm-config --libs x86codegen says:
-lLLVMX86CodeGen -lLLVMX86Info -lLLVMSelectionDAG -lLLVMAsmPrinter -lLLVMMC
-lLLVMCodeGen -lLLVMScalarOpts -lLLVMTransformUtils -lLLVMipa -lLLVMAnalysis
-lLLVMTarget -lLLVMCore -lLLVMSupport -lLLVMSystem

Looking at the -why output from GenDepLibs.pl I see:
libLLVMSelectionDAG.a:
 libLLVMAsmPrinter.a
  _ZN4llvm11DwarfWriter15RecordRegionEndEPNS_14GlobalVariableE
  _ZN4llvm11DwarfWriter16RecordSourceLineEjjNS_13DICompileUnitE
  _ZN4llvm11DwarfWriter17RecordRegionStartEPNS_14GlobalVariableE
  _ZN4llvm11DwarfWriter18RecordInlinedFnEndENS_12DISubprogramE
_ZN4llvm11DwarfWriter19RecordVariableScopeERNS_10DIVariableEPKNS_12MachineInstrE
 
_ZN4llvm11DwarfWriter20RecordInlinedFnStartENS_12DISubprogramENS_13DICompileUnitEjj
  _ZN4llvm11DwarfWriter2IDE
  _ZNK4llvm11DwarfWriter20ShouldEmitDwarfDebugEv

See attached full -why output.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 17:01:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 17:01:37 -0500
Subject: [LLVMbugs] [Bug 4645] crash in simplify libcalls
In-Reply-To: <bug-4645-206@http.llvm.org/bugs/>
Message-ID: <200907292201.n6TM1beh011508@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4645


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 29 18:44:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 29 Jul 2009 18:44:23 -0500
Subject: [LLVMbugs] [Bug 4646] Key manager, EH frames,
 JIT and Mac OS X 10.5.7 (llvm 2.5)
In-Reply-To: <bug-4646-206@http.llvm.org/bugs/>
Message-ID: <200907292344.n6TNiNBi015893@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4646


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-29 18:44:23 ---
This was fixed in r69314, which is in mainline but came after 2.5.  You can
probably pull that one patch into your tree for a fix or use mainline.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 12:06:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 12:06:05 -0500
Subject: [LLVMbugs] [Bug 4650] New: weak_odr symbol has "W" modifier
Message-ID: <bug-4650-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4650

           Summary: weak_odr symbol has "W" modifier
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rafael.espindola at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Compiling
-----------------------------------------------------------
target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32"
target triple = "i386-unknown-linux-gnu"
@foo = weak_odr constant [1 x i8] c"\01"
------------------------------------------------------------

produces
        .section        .gnu.linkonce.r.foo,"aM", at progbits

Before revision 77184 it would produce

.section        .gnu.linkonce.r.foo,"a", at progbits

This causes the gnu assembler to produce the warning:

Warning: entity size for SHF_MERGE not specified


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 13:38:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 13:38:00 -0500
Subject: [LLVMbugs] [Bug 4651] New: MMX testcase ICEs llvm-gcc in Darwin/x86
Message-ID: <bug-4651-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4651

           Summary: MMX testcase ICEs llvm-gcc in Darwin/x86
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: Mac System 9.x
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: stuart at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3260)
 --> (http://llvm.org/bugs/attachment.cgi?id=3260)
trivial MMX testcase reduced from some GCC DejaGNU test

On Darwin/x86:

[citroen:/Volumes/sandbox/stuart/llvm-gcc.pure.debug.obj] hasting2%
/usr/local/test/bin/llvm-gcc -S /tmp/mmx1.c
Assertion failed: ((Result == 0 || VOID_TYPE_P(TREE_TYPE(exp)) ||
Result->getType() == ConvertType(TREE_TYPE(exp))) && "Value has wrong type!"),
f
unction Emit, file ../../llvm-gcc.pure/gcc/llvm-convert.cpp, line 967.
/tmp/mmx1.c: In function 't1':
/tmp/mmx1.c:5: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 13:41:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 13:41:25 -0500
Subject: [LLVMbugs] [Bug 4652] New: Regression: null pointers no longer
 generate warnings
Message-ID: <bug-4652-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4652

           Summary: Regression: null pointers no longer generate warnings
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: Mac System 9.x
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: stuart at apple.com
                CC: llvmbugs at cs.uiuc.edu





-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 16:34:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 16:34:14 -0500
Subject: [LLVMbugs] [Bug 4654] New: Assertion failed: (V ==
 Op.getLiteralValue() && " Invalid abbrev for record!") when writing .bc
 file with metadata
Message-ID: <bug-4654-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4654

           Summary: Assertion failed: (V == Op.getLiteralValue() && "Invalid
                    abbrev for record!") when writing .bc file with metadata
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Bitcode Writer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: benny.kra at gmail.com
                CC: dpatel at apple.com, llvmbugs at cs.uiuc.edu


$ cat t.ll
@foo = constant i1 false

!a= !{!0}
!0 = metadata !{i1 false}

$ llvm-as t.ll
Assertion failed: (V == Op.getLiteralValue() && "Invalid abbrev for record!"),
function EmitAbbreviatedLiteral, file
/Users/d0k/Desktop/build/llvm/include/llvm/Bitcode/BitstreamWriter.h, line 266.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 18:07:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 18:07:44 -0500
Subject: [LLVMbugs] [Bug 4654] Assertion failed: (V == Op.getLiteralValue()
 && " Invalid abbrev for record!") when writing .bc file with metadata
In-Reply-To: <bug-4654-206@http.llvm.org/bugs/>
Message-ID: <200907302307.n6UN7i8M021781@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4654


devang.patel <devang.patel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|dpatel at apple.com            |devang.patel at gmail.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from devang.patel <devang.patel at gmail.com>  2009-07-30 18:07:43 ---
Fixed. Rev 77637.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 30 19:51:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 30 Jul 2009 19:51:29 -0500
Subject: [LLVMbugs] [Bug 4657] New: register scavenger asserts with subreg
	lowering
Message-ID: <bug-4657-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4657

           Summary: register scavenger asserts with subreg lowering
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Register Allocator
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bob.wilson at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3262)
 --> (http://llvm.org/bugs/attachment.cgi?id=3262)
testcase

The attached file causes the scavenger to assert when compiled with "llc
-march=arm -mattr=+neon".  Here is what I think is happening.  (I've actually
been looking at a slightly different case that doesn't occur without some of my
local changes, but I think it is the same problem.)  A Neon quad register is
defined by a pair of insert_subregs.  The subreg accesses are lowered into
copies that assign the underlying Neon double registers.  Now you've got a quad
register that is used but never (directly) defined.  Or, at least that seems to
be what the scavenger is thinking.

For example, in the following Q1 is defined:

        %D0<def> = FMDRR %R2<kill>, %R3<kill>, 14, %reg0
        %D1<def> = FMDRR %R0<kill>, %R1<kill>, 14, %reg0
        %Q1<def> = INSERT_SUBREG %Q1<undef>, %D1<kill>, 5
        %Q1<def> = INSERT_SUBREG %Q1, %D0<kill>, 6
        %R0<def> = ADDri <fi#0>, 0, 14, %reg0, %reg0
        VSTRQ %Q1<kill>, %R0, Mem:ST(16,16) [v_addr + 0]

but then the INSERT_SUBREGs are lowered:

********** LOWERING SUBREG INSTRS **********
********** Function: scale
subreg: CONVERTING: %Q1<def> = INSERT_SUBREG %Q1<undef>, %D1<kill>, 5
subreg: %D2<def> = FCPYD %D1, 14, %reg0
subreg: CONVERTING: %Q1<def> = INSERT_SUBREG %Q1, %D0<kill>, 6subreg: %D3<def>
= FCPYD %D0, 14, %reg0

and the scavenger asserts:

Assertion failed: (isUsed(Reg) && "Using an undefined register!"), function
forward, file
/Users/bwilson/local/llvm/llvm/lib/CodeGen/RegisterScavenging.cpp, line 221.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 00:13:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 00:13:34 -0500
Subject: [LLVMbugs] [Bug 4659] New: thumb-2 emits invalid instruction sub.w
	sp, rX, #xx
Message-ID: <bug-4659-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4659

           Summary: thumb-2 emits invalid instruction sub.w sp,rX,#xx
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: deeppatel1987 at gmail.com
                CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu


Created an attachment (id=3265)
 --> (http://llvm.org/bugs/attachment.cgi?id=3265)
reduction, but not very small

The attached reduction shows Thumb-2 emitting invalid instructions.

If you're using gas, you may need the attached patch to get to the bad "sub.w
sp,r7,#12" error this PR is about.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 03:58:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 03:58:21 -0500
Subject: [LLVMbugs] [Bug 4661] New: Assertion `(Ty == 0 || Ty ==
 V->getType()) && " Type mismatch in value table!"' failed. On file with
 named metadata
Message-ID: <bug-4661-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4661

           Summary: Assertion `(Ty == 0 || Ty == V->getType()) && "Type
                    mismatch in value table!"' failed. On file with named
                    metadata
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: tomas.l.olsen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3267)
 --> (http://llvm.org/bugs/attachment.cgi?id=3267)
bitcode to reproduce assertion failure

I'm attaching the full bitcode of the D module compiled with LDC, since
bugpoint fails with the same assertion.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 04:22:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 04:22:54 -0500
Subject: [LLVMbugs] [Bug 4662] New: Assertion `Traits::getNext(NodePtr) != 0
 && " Dereferencing end()!"' failed. with -tailduplicate
Message-ID: <bug-4662-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4662

           Summary: Assertion `Traits::getNext(NodePtr) != 0 &&
                    "Dereferencing end()!"' failed. with -tailduplicate
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: regression
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: tomas.l.olsen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3268)
 --> (http://llvm.org/bugs/attachment.cgi?id=3268)
bitcode that shows -tailduplicate problem

This seems to be a regression, attaching bitcode to reproduce like so:

opt foo.bc -tailduplicate -o=fooopt.bc -f


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 11:18:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 11:18:00 -0500
Subject: [LLVMbugs] [Bug 4650] size missing in mergeable section
In-Reply-To: <bug-4650-206@http.llvm.org/bugs/>
Message-ID: <200907311618.n6VGI0VE008529@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4650


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Chris Lattner <clattner at apple.com>  2009-07-31 11:17:59 ---
I opted to go for the simple fix:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090727/083081.html

This returns us back to previous behavior.  I'd like to get further along with
the section selection refactoring stuff before making it more complex.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 14:38:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 14:38:57 -0500
Subject: [LLVMbugs] [Bug 4662] Assertion `Traits::getNext(NodePtr) != 0 &&
 "Dereferencing end() !"' failed. with -tailduplicate
In-Reply-To: <bug-4662-206@http.llvm.org/bugs/>
Message-ID: <200907311938.n6VJcv71016244@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4662


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-07-31 14:38:57 ---
Fixed in r77716.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 15:33:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 15:33:32 -0500
Subject: [LLVMbugs] [Bug 4664] New: instcombine generates invalid constant
	operands
Message-ID: <bug-4664-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4664

           Summary: instcombine generates invalid constant operands
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: kevin.fan at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3270)
 --> (http://llvm.org/bugs/attachment.cgi?id=3270)
input bitcode file

Instcombine generates invalid constant operands in the presence of certain
sext/trunc instructions.

To reproduce:

$ opt -inline -instcombine bug.bc -f -o out.bc

Observe that the disassembly file out.ll contains bogus constants, for example:
   %0 = malloc [3758096388 x i8*]

In addition, if you try to run llc on the output:

$ llc out.bc
llc: /home/kevin/llvm/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:872:
llvm::SDValue llvm::SelectionDAG::getConstant(uint64_t, llvm::MVT, bool):
Assertion `(EltVT.getSizeInBits() >= 64 || (uint64_t)((int64_t)Val >>
EltVT.getSizeInBits()) + 1 < 2) && "getConstant with a uint64_t value that
doesn't fit in the type!"' failed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 31 15:59:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 31 Jul 2009 15:59:35 -0500
Subject: [LLVMbugs] [Bug 4664] instcombine generates invalid constant
	operands
In-Reply-To: <bug-4664-206@http.llvm.org/bugs/>
Message-ID: <200907312059.n6VKxZd3019423@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4664


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-07-31 15:59:34 ---
This will magically be fixed when victor rips out the malloc instruction. 
Sorry :(

*** This bug has been marked as a duplicate of bug 715 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.